University of Birmingham

Optimising host cell physiology and stressavoidance for the production of recombinant humantumour necrosis factor in Escherichia coliSelas Castineiras Tania Williams Steven Hitchcock Antony Cole Jeffrey Smith DanielOverton TimDOI101099mic0000622

Document VersionEarly version also known as pre-print

Citation for published version (Harvard)Selas Castineiras T Williams S Hitchcock A Cole J Smith D amp Overton T 2018 Optimising host cellphysiology and stress avoidance for the production of recombinant human tumour necrosis factor in Escherichiacoli Microbiology vol 164 pp 440-452 httpsdoiorg101099mic0000622

Link to publication on Research at Birmingham portal

General rightsUnless a licence is specified above all rights (including copyright and moral rights) in this document are retained by the authors andor thecopyright holders The express permission of the copyright holder must be obtained for any use of this material other than for purposespermitted by law

bullUsers may freely distribute the URL that is used to identify this publicationbullUsers may download andor print one copy of the publication from the University of Birmingham research portal for the purpose of privatestudy or non-commercial researchbullUser may use extracts from the document in line with the concept of lsquofair dealingrsquo under the Copyright Designs and Patents Act 1988 ()bullUsers may not further distribute the material nor use it for the purposes of commercial gain

Where a licence is displayed above please note the terms and conditions of the licence govern your use of this document

When citing please reference the published version

Take down policyWhile the University of Birmingham exercises care and attention in making items available there are rare occasions when an item has beenuploaded in error or has been deemed to be commercially or otherwise sensitive

If you believe that this is the case for this document please contact UBIRAlistsbhamacuk providing details and we will remove access tothe work immediately and investigate

Download date 14 Jan 2021

1

PREPRINT 1

Optimising host cell physiology and stress avoidance for the production of recombinant 2

human tumour necrosis factor α in Escherichia coli 3

Tania Selas Castintildeeiras123 Steven G Williams1 Antony Hitchcock1 Jeffrey A Cole34 Daniel C 4

Smith1 Tim W Overton23 5

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 6

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 7

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 8

To whom correspondence should be sent twovertonbhamacuk +44 (0) 121 414 5306 9

Keywords Heterologous protein High Cell Density Culture Fed-batch fermentation Protein 10

solubility Biopharmaceutical 11

Subject category Biotechnology 12

Word count 4958 13

14

ABSTRACT 15

As high-level recombinant protein production (RPP) exerts a massive stress on the production host 16

an extensive literature on RPP optimisation focuses on separating the growth phase from RPP 17

production once sufficient biomass has been obtained The aim of the current investigation was to 18

optimise benefits of the relatively neglected alternative strategy to achieve high level RPP during 19

growth by minimizing stress on the host High yields of the biopharmaceutical recombinant human 20

Tumour Necrosis Factor alpha (rhTNFα) were obtained by fed-batch fermentation relevant to 21

industrial production based upon parameters that most severely affected RPP in preliminary 22

laboratory scale batch cultures Decreasing the inducer concentration and growth temperature but 23

2

increasing the production period were far more effective for increasing RPP yields than changing the 24

growth phase at which production was induced High yields of up to 5 gmiddotL-1 of rhTNFα were obtained 25

with minimal plasmid loss even in synthetic media that lack animal-derived components and are 26

therefore fully compliant with regulatory requirements Most of the product was soluble and 27

biologically active In summary stress minimisation was shown to be an effective way to optimise 28

production of rhTNFα Data generated in shake-flask experiments allowed design of intensified 29

bioreactor cultures in which RPP and growth could be balanced leading to higher yield of both 30

rhTNFα and biomass than previous fermentations Additional benefits of this approach include 31

avoidance of lysis during harvesting and downstream processing and the ability to adjust the process 32

to minimize the need for long periods of staff supervision 33

INTRODUCTION 34

Recombinant proteins (RP) for use as human biopharmaceuticals represent a commercially 35

important product group for the pharmaceutical industry with 40 of pharmaceutical sales 36

accounted for by biopharmaceutical drugs Seven of the top-selling 10 drug products in 2013 were 37

protein biologics [1] Bacteria such as Escherichia coli are favoured as the production host when the 38

product is relatively simple and does not require glycosylation or other extensive posttranslational 39

modification for function E coli remains an important host for biopharmaceutical production due to 40

its relative simplicity when compared to mammalian cell systems such as CHO and its ease of use 41

[2] 42

An extensive literature documents two major routes for RP production in E coli generation of RPs 43

in insoluble inclusion bodies which although easy to harvest need first to be denatured then refolded 44

in vitro to a functional soluble form [3] or generation of soluble functional RP in vivo [4] Although 45

there are industrial examples of both routes the former pathway relies upon successful refolding 46

following inclusion body denaturation which for many RPs can be very inefficient complex and 47

costly Therefore production of soluble RP in E coli remains an important objective of process 48

development 49

3

Industrially relevant fermentation processes are designed to generate large quantities of RP along 50

with high biomass yields These conflicting objectives result in severe physiological stress on the 51

bacterial host due to competition between the two processes for metabolic resources such as amino 52

acids and other metabolites aminoacylated tRNAs energy and reducing power Many successful 53

fermentations avoid this conflict by separating the growth and RP production phases thereby 54

minimizing the selection of unproductive plasmid-free bacteria or selection of mutants defective in 55

RP accumulation In the current study we have exploited a less studied approach involving 56

concomitant growth and RP production under conditions that decrease the stress on the host 57

bacteria [567] Previous studies have shown that by growing bacteria at a lower temperature and 58

inducing production at a lower level by use of weaker promoters or lower inducer concentrations RP 59

production can be more easily balanced with biomass accumulation allowing higher biomass 60

concentrations to be achieved As RP is generated more slowly successful folding is enhanced 61

thus increasing soluble protein production However only a very few of the previous studies using 62

this approach have reported the effects of all of the key variables such as the effects of medium 63

composition temperature inducer concentration the structure of the recombinant plasmid and then 64

reported process development into fed batch cultivation In many cases the target was green 65

fluorescent protein rather than an industrially important product and the medium components used 66

would not meet current GMP (Good Manufacturing Practice) requirements 67

Human Tumour Necrosis factor (TNFα) is a cell signalling protein involved in systemic inflammation 68

and its primary role is the regulation of immune cells TNFα is first synthesised in humans as a 26 69

kDa transmembrane precursor protein which is processed into an active soluble 17 kDa protein 70

that associates into homotrimers [8] Recombinant human TNFα (rhTNFα) is currently on the market 71

under the international non-proprietary name tasonermin It is expressed as the soluble 17 kDa 72

monomer in E coli It was approved by the European Medicines Agency in 1999 for the treatment of 73

soft-tissue sarcoma and commercially produced by Boehringer Ingelheim under the trade name of 74

Beromunreg TNFα was selected as a model RP in this study due to its commercial relevance because 75

it has been used for other studies as a model protein for cytoplasmic RP production in E coli [9] and 76

since reference material can be commercially obtained Our first aim was to define parameters that 77

4

are most significant for the production of soluble rhTNFα in E coli shake-flask cultures Data from 78

these initial studies were used to direct development of high cell density fed-batch bioreactor cultures 79

that lack components of animal origin The results demonstrate that stress minimisation can be 80

successfully applied to generate soluble rhTNFα production in an industrially relevant process 81

METHODS 82

Bacterial strain and plasmids 83

E coli BL21-T7 (F- ompT lon hsdSB(rB- mB

-) gal dcm araBADT7RNAP) sourced from Cobra 84

Biologics (Keele UK) was used for the production of rhTNFα The gene coding for hTNFα was 85

synthesised and cloned into the pLT72 vector (Cobra Biologics Keele UK) under the transcriptional 86

control of the T7 promoter Addition of arabinose to E coli BL21-T7 induces production of the T7 87

RNA polymerase inducing expression from pLT72 Additionally three different vectors were 88

generated containing (i) the sequence encoding the hTNFα gene and the T7 terminator sequence 89

downstream from the multiple cloning site (pLT72-T7t-TNFα) (ii) the T7 terminator sequence and 90

the kanamycin gene in reverse orientation (pLT72-T7tKan-TNFα) and (iii) the T7 and T2 terminator 91

sequences flanking the kanamycin resistance gene (pLT72-T7tKanT2t-TNFα) Plasmid vectors 92

pLT72 and the pLT72-TNFα were kindly generated by Bruce Humphrey at Cobra Biologics 93

Shake-flask growth experiments 94

In initial experiments biomass and rhTNFα production in induced and non-induced conditions in 95

shake-flasks were compared using two commonly used media Luria Bertani broth (LB) and Terrific 96

broth (TB) Luria Bertani (LB) agar contained 10 gmiddotL-1 BBLTM phytone peptone (BD) 5 gmiddotL-1 BactoTM 97

yeast extract (BD) 5 gmiddotL-1 NaCl and 15 gmiddotL- extra-pure agar (Merck Millipore) in deionised water 98

Luria Bertani (LB) broth contained 10 gmiddotL-1 BBLTM phytone peptone 5 gmiddotL-1 BactoTM yeast extract 99

and 5 gmiddotL-1 NaCl in deionised water Terrific broth (Life technologies) contained 47 gmiddotL-1 of premade 100

terrific broth powder (equivalent to 118 gmiddotL-1 SELECT peptone 140 236 gmiddotL-1 yeast extract 94 gmiddotL-101

1 K2HPO4 and 22 gmiddotL-1 KH2PO4) and 4 mLmiddotL-1 of glycerol in deionised water Starter cultures were 102

grown overnight at 30 degC and 200 rpm from a single colony of bacteria in 10 mL of LB broth with 50 103

5

μgmiddotmL-1 kanamycin in a 20 mL bottle Cultures were grown in 50 mL of LB or TB supplemented with 104

50 μgmiddotmL-1 kanamycin in 250 mL baffled shake-flasks Sufficient inoculum was added to achieve a 105

starting OD600 of 01 Upon induction casamino acids were added to cultures to a final concentration 106

of 2 as it has been reported that the addition of casamino acids has a beneficial effect on rhTNFα 107

production [10] 108

Fed-batch fermentation methods are described in Supplemental information 109

For harvest of cell pellets for purification of rhTNFα the culture was centrifuged at 3500 g at 4 degC 110

for 30 min (Sorvall RC3B Plus rotor Sorvall HLR6H6000AHBBC) Pellets were resuspended in 111

phosphate buffered saline (PBS Gibco Life Technologies) and homogenised using a Dounce 112

homogeniser The homogenised cell paste was centrifuged at 7500 g at 4 degC for 30 min (Sorvall 113

RC53 Plus rotor Sorvall SS-34) Cell paste was stored at -20 degC 114

Analysis techniques 115

The optical density of cultures at 600 nm (OD600) was measured using an Amersham Pharmacia 116

Ultrospec 1100 Pro UV Visible Spectrophotometer Culture samples were also serially diluted in 117

PBS and plated onto LB agar for determination of CFU For plasmid retention analysis LB agar 118

plates were incubated at 37 degC overnight colonies were transferred by replica plating to LB agar and 119

LB agar supplemented with 50 mgmiddotL-1 kanamycin and incubated overnight at 37 degC 120

Subcellular fractionation 121

For separation of soluble and insoluble protein fractions a volume of culture equivalent to 1 mL at 122

an OD600 of 1 was centrifuged at 12000 g for 10 min Pellets were re-suspended in 250 microL of 50 mM 123

Tris-HCI pH 8 10 mM MgCl2 and 1 microL of benzonase nuclease (Merck Millipore) and incubated on 124

ice Lysozyme (3 microL of 10 mgmiddotmL-1 Sigma-Aldrich) was added and samples incubated on ice for 30 125

mins Cells were lysed using freeze thaw cycles a minimum of 3 cycles of freeze (ethanoldry ice 126

bath) and thaw (37 degC) were carried out for each sample Samples were centrifuged at 12000 g for 127

30 min to separate the soluble (supernatant) and insoluble (pellet) protein fractions Pellets were re-128

6

suspended in 250 microL of 50 mM Tris-HCI pH 8 10 mM MgCl2 constituting the insoluble protein 129

fraction 130

SDS-PAGE 131

SDS-PAGE and Western blotting were performed according to standard methods as detailed in the 132

supplemental information 133

Purification of rhTNFα 134

The purification of rhTNFα obtained from fermentation studies was carried out by Nicola Barison at 135

Cobra Biologics A proprietary purification protocol was used for the purification of rhTNFα and only 136

a summary of the process will be described here The cell paste obtained from fermentation studies 137

was resuspended and cells were disrupted by the use of a high-pressure cell disruption system 138

(Constant systems) The soluble protein fraction was obtained by centrifugation and clarified 139

rhTNFα was purified by a process comprising an ammonium sulphate precipitation and several 140

chromatography steps including anion exchange and heparin affinity chromatography The final 141

product presented a purity greater than 95 as quantified by densitometry from SDS-PAGE gels 142

(Supplemental Fig S1) 143

TNFα cytotoxicity assay 144

The C3H mouse fibrosarcoma cell line L929 a cell line sensitive to the activity of TNFα was used 145

to evaluate the activity of rhTNFα produced by fed-batch fermentation [11] L929 cells were grown 146

in T225 flasks with Eaglersquos minimum essential medium (EMEM) supplemented with 10 (vv) fetal 147

bovine serum (FBS) 2 mM glutamine and 01 NaHCO3 The cell culture was incubated at 37 degC 148

and 5 CO2 for three days Once they reached confluency cell cultures were passaged to a new 149

T225 flask by removing the culture medium washing the cells with PBS trypsinisation to detach 150

cells resuspension in fresh EMEM and transfer to new T225 flasks with fresh EMEM 151

For the cytotoxic bioassay 6times104 cells suspended in EMEM medium were added to each well of a 152

96 well plate and incubated at 37 degC and 5 CO2 for 18 hours Once confluent the medium was 153

7

exchanged for fresh EMEM medium containing 2 μgmiddotmL-1 actinomycin D a cell growth inhibitor 154

preventing cell proliferation and sensitising the cells to the activity of rhTNFα leading to apoptosis 155

[11] Different concentrations of rhTNFα reference material (Life Technologies) rhTNFα purified from 156

cell paste generated in fermentation 1 or buffer were added to the plates and incubated for 12 - 18 157

hours at 37 degC and 5 CO2 After incubation the culture medium was discarded and 200 μL staining 158

solution (05 (vv) crystal violet in 20 (vv) methanol) was added to each well for 10 minutes 159

The staining solution was discarded by inversion of the plate and excess staining solution removed 160

by the addition of deionised water The waste was removed and the L929 cells were solubilised by 161

the addition of 100 μL of 1 (wv) sodium dodecyl sulphate solution The plate was incubated for 1 162

hour on a rotary shaker at 180 rpm The OD580 of each well was measured using a FLUOstar Omega 163

Microplate Reader (BMG LABTECH) The mean absorbance for each triplicate set of standards or 164

samples assayed were calculated Using the mean absorbance data the percentage of cytotoxicity 165

was calculated using equation 2 166

cytotoxicity= [1 minusabsorbance of sample

absorbance of zero TNFα control ] times 100 (2) 167

The concentration of purified rhTNFα giving rise to a cytotoxicity value of 50 (LD50) was 168

determined A hTNFα standard curve was constructed by plotting the percentage cytotoxicity values 169

for the hTNFα standards against their concentration using GraphPad Prismreg software The standard 170

curve was used to calculate the specific activity of the purified rhTNFα A concentration of 1 unit (U) 171

of specific activity per mL is defined as that which gives rise to 50 cytotoxicity in the bioassay (ie 172

the LD50) 173

RESULTS AND DISCUSSION 174

Selection of culture medium for the production of rhTNFα 175

Expression of rhTNFα was driven from an arabinose-induced T7 expression system similar to the 176

widely-used DE3 pET system [12] E coli strain BL21-T7 was used as a host carrying a 177

chromosomal T7 RNA polymerase gene under the control of an arabinose-inducible promoter The 178

gene encoding rhTNFα was cloned into pLT72 under the control of a T7-dependent promoter In 179

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

3

Industrially relevant fermentation processes are designed to generate large quantities of RP along 50

with high biomass yields These conflicting objectives result in severe physiological stress on the 51

bacterial host due to competition between the two processes for metabolic resources such as amino 52

acids and other metabolites aminoacylated tRNAs energy and reducing power Many successful 53

fermentations avoid this conflict by separating the growth and RP production phases thereby 54

minimizing the selection of unproductive plasmid-free bacteria or selection of mutants defective in 55

RP accumulation In the current study we have exploited a less studied approach involving 56

concomitant growth and RP production under conditions that decrease the stress on the host 57

bacteria [567] Previous studies have shown that by growing bacteria at a lower temperature and 58

inducing production at a lower level by use of weaker promoters or lower inducer concentrations RP 59

production can be more easily balanced with biomass accumulation allowing higher biomass 60

concentrations to be achieved As RP is generated more slowly successful folding is enhanced 61

thus increasing soluble protein production However only a very few of the previous studies using 62

this approach have reported the effects of all of the key variables such as the effects of medium 63

composition temperature inducer concentration the structure of the recombinant plasmid and then 64

reported process development into fed batch cultivation In many cases the target was green 65

fluorescent protein rather than an industrially important product and the medium components used 66

would not meet current GMP (Good Manufacturing Practice) requirements 67

Human Tumour Necrosis factor (TNFα) is a cell signalling protein involved in systemic inflammation 68

and its primary role is the regulation of immune cells TNFα is first synthesised in humans as a 26 69

kDa transmembrane precursor protein which is processed into an active soluble 17 kDa protein 70

that associates into homotrimers [8] Recombinant human TNFα (rhTNFα) is currently on the market 71

under the international non-proprietary name tasonermin It is expressed as the soluble 17 kDa 72

monomer in E coli It was approved by the European Medicines Agency in 1999 for the treatment of 73

soft-tissue sarcoma and commercially produced by Boehringer Ingelheim under the trade name of 74

Beromunreg TNFα was selected as a model RP in this study due to its commercial relevance because 75

it has been used for other studies as a model protein for cytoplasmic RP production in E coli [9] and 76

since reference material can be commercially obtained Our first aim was to define parameters that 77

4

are most significant for the production of soluble rhTNFα in E coli shake-flask cultures Data from 78

these initial studies were used to direct development of high cell density fed-batch bioreactor cultures 79

that lack components of animal origin The results demonstrate that stress minimisation can be 80

successfully applied to generate soluble rhTNFα production in an industrially relevant process 81

METHODS 82

Bacterial strain and plasmids 83

E coli BL21-T7 (F- ompT lon hsdSB(rB- mB

-) gal dcm araBADT7RNAP) sourced from Cobra 84

Biologics (Keele UK) was used for the production of rhTNFα The gene coding for hTNFα was 85

synthesised and cloned into the pLT72 vector (Cobra Biologics Keele UK) under the transcriptional 86

control of the T7 promoter Addition of arabinose to E coli BL21-T7 induces production of the T7 87

RNA polymerase inducing expression from pLT72 Additionally three different vectors were 88

generated containing (i) the sequence encoding the hTNFα gene and the T7 terminator sequence 89

downstream from the multiple cloning site (pLT72-T7t-TNFα) (ii) the T7 terminator sequence and 90

the kanamycin gene in reverse orientation (pLT72-T7tKan-TNFα) and (iii) the T7 and T2 terminator 91

sequences flanking the kanamycin resistance gene (pLT72-T7tKanT2t-TNFα) Plasmid vectors 92

pLT72 and the pLT72-TNFα were kindly generated by Bruce Humphrey at Cobra Biologics 93

Shake-flask growth experiments 94

In initial experiments biomass and rhTNFα production in induced and non-induced conditions in 95

shake-flasks were compared using two commonly used media Luria Bertani broth (LB) and Terrific 96

broth (TB) Luria Bertani (LB) agar contained 10 gmiddotL-1 BBLTM phytone peptone (BD) 5 gmiddotL-1 BactoTM 97

yeast extract (BD) 5 gmiddotL-1 NaCl and 15 gmiddotL- extra-pure agar (Merck Millipore) in deionised water 98

Luria Bertani (LB) broth contained 10 gmiddotL-1 BBLTM phytone peptone 5 gmiddotL-1 BactoTM yeast extract 99

and 5 gmiddotL-1 NaCl in deionised water Terrific broth (Life technologies) contained 47 gmiddotL-1 of premade 100

terrific broth powder (equivalent to 118 gmiddotL-1 SELECT peptone 140 236 gmiddotL-1 yeast extract 94 gmiddotL-101

1 K2HPO4 and 22 gmiddotL-1 KH2PO4) and 4 mLmiddotL-1 of glycerol in deionised water Starter cultures were 102

grown overnight at 30 degC and 200 rpm from a single colony of bacteria in 10 mL of LB broth with 50 103

5

μgmiddotmL-1 kanamycin in a 20 mL bottle Cultures were grown in 50 mL of LB or TB supplemented with 104

50 μgmiddotmL-1 kanamycin in 250 mL baffled shake-flasks Sufficient inoculum was added to achieve a 105

starting OD600 of 01 Upon induction casamino acids were added to cultures to a final concentration 106

of 2 as it has been reported that the addition of casamino acids has a beneficial effect on rhTNFα 107

production [10] 108

Fed-batch fermentation methods are described in Supplemental information 109

For harvest of cell pellets for purification of rhTNFα the culture was centrifuged at 3500 g at 4 degC 110

for 30 min (Sorvall RC3B Plus rotor Sorvall HLR6H6000AHBBC) Pellets were resuspended in 111

phosphate buffered saline (PBS Gibco Life Technologies) and homogenised using a Dounce 112

homogeniser The homogenised cell paste was centrifuged at 7500 g at 4 degC for 30 min (Sorvall 113

RC53 Plus rotor Sorvall SS-34) Cell paste was stored at -20 degC 114

Analysis techniques 115

The optical density of cultures at 600 nm (OD600) was measured using an Amersham Pharmacia 116

Ultrospec 1100 Pro UV Visible Spectrophotometer Culture samples were also serially diluted in 117

PBS and plated onto LB agar for determination of CFU For plasmid retention analysis LB agar 118

plates were incubated at 37 degC overnight colonies were transferred by replica plating to LB agar and 119

LB agar supplemented with 50 mgmiddotL-1 kanamycin and incubated overnight at 37 degC 120

Subcellular fractionation 121

For separation of soluble and insoluble protein fractions a volume of culture equivalent to 1 mL at 122

an OD600 of 1 was centrifuged at 12000 g for 10 min Pellets were re-suspended in 250 microL of 50 mM 123

Tris-HCI pH 8 10 mM MgCl2 and 1 microL of benzonase nuclease (Merck Millipore) and incubated on 124

ice Lysozyme (3 microL of 10 mgmiddotmL-1 Sigma-Aldrich) was added and samples incubated on ice for 30 125

mins Cells were lysed using freeze thaw cycles a minimum of 3 cycles of freeze (ethanoldry ice 126

bath) and thaw (37 degC) were carried out for each sample Samples were centrifuged at 12000 g for 127

30 min to separate the soluble (supernatant) and insoluble (pellet) protein fractions Pellets were re-128

6

suspended in 250 microL of 50 mM Tris-HCI pH 8 10 mM MgCl2 constituting the insoluble protein 129

fraction 130

SDS-PAGE 131

SDS-PAGE and Western blotting were performed according to standard methods as detailed in the 132

supplemental information 133

Purification of rhTNFα 134

The purification of rhTNFα obtained from fermentation studies was carried out by Nicola Barison at 135

Cobra Biologics A proprietary purification protocol was used for the purification of rhTNFα and only 136

a summary of the process will be described here The cell paste obtained from fermentation studies 137

was resuspended and cells were disrupted by the use of a high-pressure cell disruption system 138

(Constant systems) The soluble protein fraction was obtained by centrifugation and clarified 139

rhTNFα was purified by a process comprising an ammonium sulphate precipitation and several 140

chromatography steps including anion exchange and heparin affinity chromatography The final 141

product presented a purity greater than 95 as quantified by densitometry from SDS-PAGE gels 142

(Supplemental Fig S1) 143

TNFα cytotoxicity assay 144

The C3H mouse fibrosarcoma cell line L929 a cell line sensitive to the activity of TNFα was used 145

to evaluate the activity of rhTNFα produced by fed-batch fermentation [11] L929 cells were grown 146

in T225 flasks with Eaglersquos minimum essential medium (EMEM) supplemented with 10 (vv) fetal 147

bovine serum (FBS) 2 mM glutamine and 01 NaHCO3 The cell culture was incubated at 37 degC 148

and 5 CO2 for three days Once they reached confluency cell cultures were passaged to a new 149

T225 flask by removing the culture medium washing the cells with PBS trypsinisation to detach 150

cells resuspension in fresh EMEM and transfer to new T225 flasks with fresh EMEM 151

For the cytotoxic bioassay 6times104 cells suspended in EMEM medium were added to each well of a 152

96 well plate and incubated at 37 degC and 5 CO2 for 18 hours Once confluent the medium was 153

7

exchanged for fresh EMEM medium containing 2 μgmiddotmL-1 actinomycin D a cell growth inhibitor 154

preventing cell proliferation and sensitising the cells to the activity of rhTNFα leading to apoptosis 155

[11] Different concentrations of rhTNFα reference material (Life Technologies) rhTNFα purified from 156

cell paste generated in fermentation 1 or buffer were added to the plates and incubated for 12 - 18 157

hours at 37 degC and 5 CO2 After incubation the culture medium was discarded and 200 μL staining 158

solution (05 (vv) crystal violet in 20 (vv) methanol) was added to each well for 10 minutes 159

The staining solution was discarded by inversion of the plate and excess staining solution removed 160

by the addition of deionised water The waste was removed and the L929 cells were solubilised by 161

the addition of 100 μL of 1 (wv) sodium dodecyl sulphate solution The plate was incubated for 1 162

hour on a rotary shaker at 180 rpm The OD580 of each well was measured using a FLUOstar Omega 163

Microplate Reader (BMG LABTECH) The mean absorbance for each triplicate set of standards or 164

samples assayed were calculated Using the mean absorbance data the percentage of cytotoxicity 165

was calculated using equation 2 166

cytotoxicity= [1 minusabsorbance of sample

absorbance of zero TNFα control ] times 100 (2) 167

The concentration of purified rhTNFα giving rise to a cytotoxicity value of 50 (LD50) was 168

determined A hTNFα standard curve was constructed by plotting the percentage cytotoxicity values 169

for the hTNFα standards against their concentration using GraphPad Prismreg software The standard 170

curve was used to calculate the specific activity of the purified rhTNFα A concentration of 1 unit (U) 171

of specific activity per mL is defined as that which gives rise to 50 cytotoxicity in the bioassay (ie 172

the LD50) 173

RESULTS AND DISCUSSION 174

Selection of culture medium for the production of rhTNFα 175

Expression of rhTNFα was driven from an arabinose-induced T7 expression system similar to the 176

widely-used DE3 pET system [12] E coli strain BL21-T7 was used as a host carrying a 177

chromosomal T7 RNA polymerase gene under the control of an arabinose-inducible promoter The 178

gene encoding rhTNFα was cloned into pLT72 under the control of a T7-dependent promoter In 179

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

4

are most significant for the production of soluble rhTNFα in E coli shake-flask cultures Data from 78

these initial studies were used to direct development of high cell density fed-batch bioreactor cultures 79

that lack components of animal origin The results demonstrate that stress minimisation can be 80

successfully applied to generate soluble rhTNFα production in an industrially relevant process 81

METHODS 82

Bacterial strain and plasmids 83

E coli BL21-T7 (F- ompT lon hsdSB(rB- mB

-) gal dcm araBADT7RNAP) sourced from Cobra 84

Biologics (Keele UK) was used for the production of rhTNFα The gene coding for hTNFα was 85

synthesised and cloned into the pLT72 vector (Cobra Biologics Keele UK) under the transcriptional 86

control of the T7 promoter Addition of arabinose to E coli BL21-T7 induces production of the T7 87

RNA polymerase inducing expression from pLT72 Additionally three different vectors were 88

generated containing (i) the sequence encoding the hTNFα gene and the T7 terminator sequence 89

downstream from the multiple cloning site (pLT72-T7t-TNFα) (ii) the T7 terminator sequence and 90

the kanamycin gene in reverse orientation (pLT72-T7tKan-TNFα) and (iii) the T7 and T2 terminator 91

sequences flanking the kanamycin resistance gene (pLT72-T7tKanT2t-TNFα) Plasmid vectors 92

pLT72 and the pLT72-TNFα were kindly generated by Bruce Humphrey at Cobra Biologics 93

Shake-flask growth experiments 94

In initial experiments biomass and rhTNFα production in induced and non-induced conditions in 95

shake-flasks were compared using two commonly used media Luria Bertani broth (LB) and Terrific 96

broth (TB) Luria Bertani (LB) agar contained 10 gmiddotL-1 BBLTM phytone peptone (BD) 5 gmiddotL-1 BactoTM 97

yeast extract (BD) 5 gmiddotL-1 NaCl and 15 gmiddotL- extra-pure agar (Merck Millipore) in deionised water 98

Luria Bertani (LB) broth contained 10 gmiddotL-1 BBLTM phytone peptone 5 gmiddotL-1 BactoTM yeast extract 99

and 5 gmiddotL-1 NaCl in deionised water Terrific broth (Life technologies) contained 47 gmiddotL-1 of premade 100

terrific broth powder (equivalent to 118 gmiddotL-1 SELECT peptone 140 236 gmiddotL-1 yeast extract 94 gmiddotL-101

1 K2HPO4 and 22 gmiddotL-1 KH2PO4) and 4 mLmiddotL-1 of glycerol in deionised water Starter cultures were 102

grown overnight at 30 degC and 200 rpm from a single colony of bacteria in 10 mL of LB broth with 50 103

5

μgmiddotmL-1 kanamycin in a 20 mL bottle Cultures were grown in 50 mL of LB or TB supplemented with 104

50 μgmiddotmL-1 kanamycin in 250 mL baffled shake-flasks Sufficient inoculum was added to achieve a 105

starting OD600 of 01 Upon induction casamino acids were added to cultures to a final concentration 106

of 2 as it has been reported that the addition of casamino acids has a beneficial effect on rhTNFα 107

production [10] 108

Fed-batch fermentation methods are described in Supplemental information 109

For harvest of cell pellets for purification of rhTNFα the culture was centrifuged at 3500 g at 4 degC 110

for 30 min (Sorvall RC3B Plus rotor Sorvall HLR6H6000AHBBC) Pellets were resuspended in 111

phosphate buffered saline (PBS Gibco Life Technologies) and homogenised using a Dounce 112

homogeniser The homogenised cell paste was centrifuged at 7500 g at 4 degC for 30 min (Sorvall 113

RC53 Plus rotor Sorvall SS-34) Cell paste was stored at -20 degC 114

Analysis techniques 115

The optical density of cultures at 600 nm (OD600) was measured using an Amersham Pharmacia 116

Ultrospec 1100 Pro UV Visible Spectrophotometer Culture samples were also serially diluted in 117

PBS and plated onto LB agar for determination of CFU For plasmid retention analysis LB agar 118

plates were incubated at 37 degC overnight colonies were transferred by replica plating to LB agar and 119

LB agar supplemented with 50 mgmiddotL-1 kanamycin and incubated overnight at 37 degC 120

Subcellular fractionation 121

For separation of soluble and insoluble protein fractions a volume of culture equivalent to 1 mL at 122

an OD600 of 1 was centrifuged at 12000 g for 10 min Pellets were re-suspended in 250 microL of 50 mM 123

Tris-HCI pH 8 10 mM MgCl2 and 1 microL of benzonase nuclease (Merck Millipore) and incubated on 124

ice Lysozyme (3 microL of 10 mgmiddotmL-1 Sigma-Aldrich) was added and samples incubated on ice for 30 125

mins Cells were lysed using freeze thaw cycles a minimum of 3 cycles of freeze (ethanoldry ice 126

bath) and thaw (37 degC) were carried out for each sample Samples were centrifuged at 12000 g for 127

30 min to separate the soluble (supernatant) and insoluble (pellet) protein fractions Pellets were re-128

6

suspended in 250 microL of 50 mM Tris-HCI pH 8 10 mM MgCl2 constituting the insoluble protein 129

fraction 130

SDS-PAGE 131

SDS-PAGE and Western blotting were performed according to standard methods as detailed in the 132

supplemental information 133

Purification of rhTNFα 134

The purification of rhTNFα obtained from fermentation studies was carried out by Nicola Barison at 135

Cobra Biologics A proprietary purification protocol was used for the purification of rhTNFα and only 136

a summary of the process will be described here The cell paste obtained from fermentation studies 137

was resuspended and cells were disrupted by the use of a high-pressure cell disruption system 138

(Constant systems) The soluble protein fraction was obtained by centrifugation and clarified 139

rhTNFα was purified by a process comprising an ammonium sulphate precipitation and several 140

chromatography steps including anion exchange and heparin affinity chromatography The final 141

product presented a purity greater than 95 as quantified by densitometry from SDS-PAGE gels 142

(Supplemental Fig S1) 143

TNFα cytotoxicity assay 144

The C3H mouse fibrosarcoma cell line L929 a cell line sensitive to the activity of TNFα was used 145

to evaluate the activity of rhTNFα produced by fed-batch fermentation [11] L929 cells were grown 146

in T225 flasks with Eaglersquos minimum essential medium (EMEM) supplemented with 10 (vv) fetal 147

bovine serum (FBS) 2 mM glutamine and 01 NaHCO3 The cell culture was incubated at 37 degC 148

and 5 CO2 for three days Once they reached confluency cell cultures were passaged to a new 149

T225 flask by removing the culture medium washing the cells with PBS trypsinisation to detach 150

cells resuspension in fresh EMEM and transfer to new T225 flasks with fresh EMEM 151

For the cytotoxic bioassay 6times104 cells suspended in EMEM medium were added to each well of a 152

96 well plate and incubated at 37 degC and 5 CO2 for 18 hours Once confluent the medium was 153

7

exchanged for fresh EMEM medium containing 2 μgmiddotmL-1 actinomycin D a cell growth inhibitor 154

preventing cell proliferation and sensitising the cells to the activity of rhTNFα leading to apoptosis 155

[11] Different concentrations of rhTNFα reference material (Life Technologies) rhTNFα purified from 156

cell paste generated in fermentation 1 or buffer were added to the plates and incubated for 12 - 18 157

hours at 37 degC and 5 CO2 After incubation the culture medium was discarded and 200 μL staining 158

solution (05 (vv) crystal violet in 20 (vv) methanol) was added to each well for 10 minutes 159

The staining solution was discarded by inversion of the plate and excess staining solution removed 160

by the addition of deionised water The waste was removed and the L929 cells were solubilised by 161

the addition of 100 μL of 1 (wv) sodium dodecyl sulphate solution The plate was incubated for 1 162

hour on a rotary shaker at 180 rpm The OD580 of each well was measured using a FLUOstar Omega 163

Microplate Reader (BMG LABTECH) The mean absorbance for each triplicate set of standards or 164

samples assayed were calculated Using the mean absorbance data the percentage of cytotoxicity 165

was calculated using equation 2 166

cytotoxicity= [1 minusabsorbance of sample

absorbance of zero TNFα control ] times 100 (2) 167

The concentration of purified rhTNFα giving rise to a cytotoxicity value of 50 (LD50) was 168

determined A hTNFα standard curve was constructed by plotting the percentage cytotoxicity values 169

for the hTNFα standards against their concentration using GraphPad Prismreg software The standard 170

curve was used to calculate the specific activity of the purified rhTNFα A concentration of 1 unit (U) 171

of specific activity per mL is defined as that which gives rise to 50 cytotoxicity in the bioassay (ie 172

the LD50) 173

RESULTS AND DISCUSSION 174

Selection of culture medium for the production of rhTNFα 175

Expression of rhTNFα was driven from an arabinose-induced T7 expression system similar to the 176

widely-used DE3 pET system [12] E coli strain BL21-T7 was used as a host carrying a 177

chromosomal T7 RNA polymerase gene under the control of an arabinose-inducible promoter The 178

gene encoding rhTNFα was cloned into pLT72 under the control of a T7-dependent promoter In 179

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

5

μgmiddotmL-1 kanamycin in a 20 mL bottle Cultures were grown in 50 mL of LB or TB supplemented with 104

50 μgmiddotmL-1 kanamycin in 250 mL baffled shake-flasks Sufficient inoculum was added to achieve a 105

starting OD600 of 01 Upon induction casamino acids were added to cultures to a final concentration 106

of 2 as it has been reported that the addition of casamino acids has a beneficial effect on rhTNFα 107

production [10] 108

Fed-batch fermentation methods are described in Supplemental information 109

For harvest of cell pellets for purification of rhTNFα the culture was centrifuged at 3500 g at 4 degC 110

for 30 min (Sorvall RC3B Plus rotor Sorvall HLR6H6000AHBBC) Pellets were resuspended in 111

phosphate buffered saline (PBS Gibco Life Technologies) and homogenised using a Dounce 112

homogeniser The homogenised cell paste was centrifuged at 7500 g at 4 degC for 30 min (Sorvall 113

RC53 Plus rotor Sorvall SS-34) Cell paste was stored at -20 degC 114

Analysis techniques 115

The optical density of cultures at 600 nm (OD600) was measured using an Amersham Pharmacia 116

Ultrospec 1100 Pro UV Visible Spectrophotometer Culture samples were also serially diluted in 117

PBS and plated onto LB agar for determination of CFU For plasmid retention analysis LB agar 118

plates were incubated at 37 degC overnight colonies were transferred by replica plating to LB agar and 119

LB agar supplemented with 50 mgmiddotL-1 kanamycin and incubated overnight at 37 degC 120

Subcellular fractionation 121

For separation of soluble and insoluble protein fractions a volume of culture equivalent to 1 mL at 122

an OD600 of 1 was centrifuged at 12000 g for 10 min Pellets were re-suspended in 250 microL of 50 mM 123

Tris-HCI pH 8 10 mM MgCl2 and 1 microL of benzonase nuclease (Merck Millipore) and incubated on 124

ice Lysozyme (3 microL of 10 mgmiddotmL-1 Sigma-Aldrich) was added and samples incubated on ice for 30 125

mins Cells were lysed using freeze thaw cycles a minimum of 3 cycles of freeze (ethanoldry ice 126

bath) and thaw (37 degC) were carried out for each sample Samples were centrifuged at 12000 g for 127

30 min to separate the soluble (supernatant) and insoluble (pellet) protein fractions Pellets were re-128

6

suspended in 250 microL of 50 mM Tris-HCI pH 8 10 mM MgCl2 constituting the insoluble protein 129

fraction 130

SDS-PAGE 131

SDS-PAGE and Western blotting were performed according to standard methods as detailed in the 132

supplemental information 133

Purification of rhTNFα 134

The purification of rhTNFα obtained from fermentation studies was carried out by Nicola Barison at 135

Cobra Biologics A proprietary purification protocol was used for the purification of rhTNFα and only 136

a summary of the process will be described here The cell paste obtained from fermentation studies 137

was resuspended and cells were disrupted by the use of a high-pressure cell disruption system 138

(Constant systems) The soluble protein fraction was obtained by centrifugation and clarified 139

rhTNFα was purified by a process comprising an ammonium sulphate precipitation and several 140

chromatography steps including anion exchange and heparin affinity chromatography The final 141

product presented a purity greater than 95 as quantified by densitometry from SDS-PAGE gels 142

(Supplemental Fig S1) 143

TNFα cytotoxicity assay 144

The C3H mouse fibrosarcoma cell line L929 a cell line sensitive to the activity of TNFα was used 145

to evaluate the activity of rhTNFα produced by fed-batch fermentation [11] L929 cells were grown 146

in T225 flasks with Eaglersquos minimum essential medium (EMEM) supplemented with 10 (vv) fetal 147

bovine serum (FBS) 2 mM glutamine and 01 NaHCO3 The cell culture was incubated at 37 degC 148

and 5 CO2 for three days Once they reached confluency cell cultures were passaged to a new 149

T225 flask by removing the culture medium washing the cells with PBS trypsinisation to detach 150

cells resuspension in fresh EMEM and transfer to new T225 flasks with fresh EMEM 151

For the cytotoxic bioassay 6times104 cells suspended in EMEM medium were added to each well of a 152

96 well plate and incubated at 37 degC and 5 CO2 for 18 hours Once confluent the medium was 153

7

exchanged for fresh EMEM medium containing 2 μgmiddotmL-1 actinomycin D a cell growth inhibitor 154

preventing cell proliferation and sensitising the cells to the activity of rhTNFα leading to apoptosis 155

[11] Different concentrations of rhTNFα reference material (Life Technologies) rhTNFα purified from 156

cell paste generated in fermentation 1 or buffer were added to the plates and incubated for 12 - 18 157

hours at 37 degC and 5 CO2 After incubation the culture medium was discarded and 200 μL staining 158

solution (05 (vv) crystal violet in 20 (vv) methanol) was added to each well for 10 minutes 159

The staining solution was discarded by inversion of the plate and excess staining solution removed 160

by the addition of deionised water The waste was removed and the L929 cells were solubilised by 161

the addition of 100 μL of 1 (wv) sodium dodecyl sulphate solution The plate was incubated for 1 162

hour on a rotary shaker at 180 rpm The OD580 of each well was measured using a FLUOstar Omega 163

Microplate Reader (BMG LABTECH) The mean absorbance for each triplicate set of standards or 164

samples assayed were calculated Using the mean absorbance data the percentage of cytotoxicity 165

was calculated using equation 2 166

cytotoxicity= [1 minusabsorbance of sample

absorbance of zero TNFα control ] times 100 (2) 167

The concentration of purified rhTNFα giving rise to a cytotoxicity value of 50 (LD50) was 168

determined A hTNFα standard curve was constructed by plotting the percentage cytotoxicity values 169

for the hTNFα standards against their concentration using GraphPad Prismreg software The standard 170

curve was used to calculate the specific activity of the purified rhTNFα A concentration of 1 unit (U) 171

of specific activity per mL is defined as that which gives rise to 50 cytotoxicity in the bioassay (ie 172

the LD50) 173

RESULTS AND DISCUSSION 174

Selection of culture medium for the production of rhTNFα 175

Expression of rhTNFα was driven from an arabinose-induced T7 expression system similar to the 176

widely-used DE3 pET system [12] E coli strain BL21-T7 was used as a host carrying a 177

chromosomal T7 RNA polymerase gene under the control of an arabinose-inducible promoter The 178

gene encoding rhTNFα was cloned into pLT72 under the control of a T7-dependent promoter In 179

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

6

suspended in 250 microL of 50 mM Tris-HCI pH 8 10 mM MgCl2 constituting the insoluble protein 129

fraction 130

SDS-PAGE 131

SDS-PAGE and Western blotting were performed according to standard methods as detailed in the 132

supplemental information 133

Purification of rhTNFα 134

The purification of rhTNFα obtained from fermentation studies was carried out by Nicola Barison at 135

Cobra Biologics A proprietary purification protocol was used for the purification of rhTNFα and only 136

a summary of the process will be described here The cell paste obtained from fermentation studies 137

was resuspended and cells were disrupted by the use of a high-pressure cell disruption system 138

(Constant systems) The soluble protein fraction was obtained by centrifugation and clarified 139

rhTNFα was purified by a process comprising an ammonium sulphate precipitation and several 140

chromatography steps including anion exchange and heparin affinity chromatography The final 141

product presented a purity greater than 95 as quantified by densitometry from SDS-PAGE gels 142

(Supplemental Fig S1) 143

TNFα cytotoxicity assay 144

The C3H mouse fibrosarcoma cell line L929 a cell line sensitive to the activity of TNFα was used 145

to evaluate the activity of rhTNFα produced by fed-batch fermentation [11] L929 cells were grown 146

in T225 flasks with Eaglersquos minimum essential medium (EMEM) supplemented with 10 (vv) fetal 147

bovine serum (FBS) 2 mM glutamine and 01 NaHCO3 The cell culture was incubated at 37 degC 148

and 5 CO2 for three days Once they reached confluency cell cultures were passaged to a new 149

T225 flask by removing the culture medium washing the cells with PBS trypsinisation to detach 150

cells resuspension in fresh EMEM and transfer to new T225 flasks with fresh EMEM 151

For the cytotoxic bioassay 6times104 cells suspended in EMEM medium were added to each well of a 152

96 well plate and incubated at 37 degC and 5 CO2 for 18 hours Once confluent the medium was 153

7

exchanged for fresh EMEM medium containing 2 μgmiddotmL-1 actinomycin D a cell growth inhibitor 154

preventing cell proliferation and sensitising the cells to the activity of rhTNFα leading to apoptosis 155

[11] Different concentrations of rhTNFα reference material (Life Technologies) rhTNFα purified from 156

cell paste generated in fermentation 1 or buffer were added to the plates and incubated for 12 - 18 157

hours at 37 degC and 5 CO2 After incubation the culture medium was discarded and 200 μL staining 158

solution (05 (vv) crystal violet in 20 (vv) methanol) was added to each well for 10 minutes 159

The staining solution was discarded by inversion of the plate and excess staining solution removed 160

by the addition of deionised water The waste was removed and the L929 cells were solubilised by 161

the addition of 100 μL of 1 (wv) sodium dodecyl sulphate solution The plate was incubated for 1 162

hour on a rotary shaker at 180 rpm The OD580 of each well was measured using a FLUOstar Omega 163

Microplate Reader (BMG LABTECH) The mean absorbance for each triplicate set of standards or 164

samples assayed were calculated Using the mean absorbance data the percentage of cytotoxicity 165

was calculated using equation 2 166

cytotoxicity= [1 minusabsorbance of sample

absorbance of zero TNFα control ] times 100 (2) 167

The concentration of purified rhTNFα giving rise to a cytotoxicity value of 50 (LD50) was 168

determined A hTNFα standard curve was constructed by plotting the percentage cytotoxicity values 169

for the hTNFα standards against their concentration using GraphPad Prismreg software The standard 170

curve was used to calculate the specific activity of the purified rhTNFα A concentration of 1 unit (U) 171

of specific activity per mL is defined as that which gives rise to 50 cytotoxicity in the bioassay (ie 172

the LD50) 173

RESULTS AND DISCUSSION 174

Selection of culture medium for the production of rhTNFα 175

Expression of rhTNFα was driven from an arabinose-induced T7 expression system similar to the 176

widely-used DE3 pET system [12] E coli strain BL21-T7 was used as a host carrying a 177

chromosomal T7 RNA polymerase gene under the control of an arabinose-inducible promoter The 178

gene encoding rhTNFα was cloned into pLT72 under the control of a T7-dependent promoter In 179

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

7

exchanged for fresh EMEM medium containing 2 μgmiddotmL-1 actinomycin D a cell growth inhibitor 154

preventing cell proliferation and sensitising the cells to the activity of rhTNFα leading to apoptosis 155

[11] Different concentrations of rhTNFα reference material (Life Technologies) rhTNFα purified from 156

cell paste generated in fermentation 1 or buffer were added to the plates and incubated for 12 - 18 157

hours at 37 degC and 5 CO2 After incubation the culture medium was discarded and 200 μL staining 158

solution (05 (vv) crystal violet in 20 (vv) methanol) was added to each well for 10 minutes 159

The staining solution was discarded by inversion of the plate and excess staining solution removed 160

by the addition of deionised water The waste was removed and the L929 cells were solubilised by 161

the addition of 100 μL of 1 (wv) sodium dodecyl sulphate solution The plate was incubated for 1 162

hour on a rotary shaker at 180 rpm The OD580 of each well was measured using a FLUOstar Omega 163

Microplate Reader (BMG LABTECH) The mean absorbance for each triplicate set of standards or 164

samples assayed were calculated Using the mean absorbance data the percentage of cytotoxicity 165

was calculated using equation 2 166

cytotoxicity= [1 minusabsorbance of sample

absorbance of zero TNFα control ] times 100 (2) 167

The concentration of purified rhTNFα giving rise to a cytotoxicity value of 50 (LD50) was 168

determined A hTNFα standard curve was constructed by plotting the percentage cytotoxicity values 169

for the hTNFα standards against their concentration using GraphPad Prismreg software The standard 170

curve was used to calculate the specific activity of the purified rhTNFα A concentration of 1 unit (U) 171

of specific activity per mL is defined as that which gives rise to 50 cytotoxicity in the bioassay (ie 172

the LD50) 173

RESULTS AND DISCUSSION 174

Selection of culture medium for the production of rhTNFα 175

Expression of rhTNFα was driven from an arabinose-induced T7 expression system similar to the 176

widely-used DE3 pET system [12] E coli strain BL21-T7 was used as a host carrying a 177

chromosomal T7 RNA polymerase gene under the control of an arabinose-inducible promoter The 178

gene encoding rhTNFα was cloned into pLT72 under the control of a T7-dependent promoter In 179

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

8

initial experiments E coli BL21-T7 transformed with either pLT72-TNFα or the empty vector (pLT72) 180

were grown with aeration in Luria Bertani broth (LB) or Terrific Broth (TB) At an OD600 of 1 half of 181

the cultures were induced with a final concentration of 02 (wv) arabinose and casamino acids 182

were added [10] Biomass accumulation culturability (colony forming units) plasmid retention and 183

protein production were analysed (Figs 1 amp 2) Growth of cultures transformed with either the control 184

plasmid or the production plasmid stopped soon after induction This was expected because 185

production of T7 RNA polymerase even without production of an RP induces stress responses in 186

E coli [13] The final biomass concentration in non-induced cultures in LB was higher than in TB 187

For each medium non-induced cultures containing the empty vector and the vector encoding 188

rhTNFα grew similarly However for induced cultures TB cultures grew faster and reached a higher 189

OD600 than LB cultures After either 2 h or 24 h post-induction induced cultures expressing rhTNFα 190

also had higher culturability in TB than in LB 191

SDS-PAGE analysis revealed that rhTNFα accumulated gradually after induction reaching a 192

maximum of 20 of the total cell protein after 24 hours of growth for both TB and LB (Fig 2) Very 193

little rhTNFα was present in cells before induction or in non-induced cells after 24 hours growth 194

revealing that this expression system offers tight regulation of RPP Bacterial pellets harvested after 195

24 hours growth were also fractionated into soluble and insoluble fractions SDS-PAGE revealed 196

that around 55 of the rhTNFα was present in the soluble fraction for both media As a result of its 197

buffering capacity and the slight improvements noted in growth and culturability TB was selected for 198

use in further experiments 199

SDS-PAGE analysis identified an unexpected protein band (~31 kDa) in induced cultures of cells 200

carrying both the empty vector and the vector encoding rhTNFα The concentration of this unknown 201

protein band accounted for up to 10 of the total cell protein (in samples from induced cultures 202

carrying the empty vector) It was hypothesised that this unknown protein could be the product of 203

the kanamycin resistance gene aminoglycoside 3 phosphotransferase (APH) which has a 204

comparable molecular weight The lack of terminator sequences downstream of the multiple cloning 205

site on the backbone of vector pLT72 may have allowed read-through by the T7 RNA polymerase 206

leading to the overproduction of APH 207

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

9

Effect of inducer concentration and induction point on rhTNFα production 208

Addition of 02 (wv) arabinose as an inducer in the previous experiment resulted in stress as 209

evidenced by growth arrest a decrease in viability and plasmid loss (potentially due to a decrease 210

in culturability of plasmid-containing productive bacteria) even in cultures containing the empty 211

vector The effect of different inducer concentrations (1 02 005 002 or 0002 212

arabinose added at an OD600 of 1) was tested Bacteria transformed with the empty vector (pLT72) 213

or the vector coding for rhTNFα (pLT72-TNFα) under non-inducing conditions were used as controls 214

(Fig 3) 215

Growth of cultures induced with 1 to 002 arabinose was arrested following arabinose addition 216

and more than 80 of the bacteria were plasmid deficient after 24 h The final biomass concentration 217

after 24 h was inversely proportional to the inducer concentration used but higher for induced 218

cultures transformed with the recombinant plasmid than the non-induced cultures or cultures 219

containing empty vector (Fig 3a) In contrast growth of cultures induced with 0002 arabinose 220

was only slightly inhibited (Fig 3a) and more than 80 of these bacteria had retained the plasmid 221

after 24 h (Fig 3b) 222

The concentration of rhTNFα in cells after 24 hours growth was similar in all cultures (Fig 3c) 223

independent of the arabinose concentration used showing that the lowest concentration of 224

arabinose 0002 was sufficient to fully induce the T7 expression system In addition the majority 225

of the rhTNFα was found to be accumulated in the soluble protein fraction independently of the 226

concentration of arabinose used to induce the cultures (data not shown) 227

The effect of changing the point of induction in shake-flasks was evaluated by inducing RP by adding 228

02 arabinose at an OD600 of 05 2 or 3 (Supplemental Fig S2) Unlike changing the inducer 229

concentration there were no large differences between cultures induced at different cell densities 230

Cultures induced at an OD600 of 05 grew more slowly after induction but reached higher cell densities 231

than those cultures induced at an OD600 of 2 or 3 (Supplemental Fig S2a) There were no significant 232

differences in CFU or in rhTNFα productivity for cultures induced at an OD600 of 05 2 or 3 and 233

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

10

plasmid loss was observed 24 hours post-inoculation for all the induced cultures as a result of the 234

high concentration of arabinose used (Supplementary Fig 2b and data not shown) 235

Effect of temperature on rhTNFα productivity 236

As well as lowering inducer concentration stress minimisation can be achieved by decreasing the 237

temperature of growth thus slowing growth and protein production rates This has been previously 238

been shown to favour accumulation of recombinant proteins in a soluble form [614] Cultures were 239

grown as before in TB supplemented with casamino acids but at temperatures of 37 degC 30 degC or 240

25 degC Expression of rhTNFα was induced by the addition of arabinose to a final concentration of 02 241

at OD600 = 1 cell samples were harvested 4 hours after induction separated into soluble and 242

insoluble fractions and analysed by SDS-PAGE (Fig 4a) The proportion of rhTNFα in the soluble 243

fraction increased as the growth temperature decreased 244

To investigate the interplay between inducer concentration and temperature of growth further 245

cultures were grown as before at 25 degC and at OD600 = 05 induced with concentrations of arabinose 246

between 02 and 0001 Plasmids were retained for 24 h by uninduced cultures or cultures 247

transformed with the empty vector (Fig 4b) Although gt90 of bacteria induced with 02 248

arabinose had lost the plasmid within 24 h plasmids had been retained by most of the bacteria at 249

much higher induced concentrations at 25 degC than at 30 degC (Compare Figs 3b and 4b) Production 250

of rhTNFα production was fully induced with 0005 and 0002 arabinose but was suboptimally 251

induced at 0001 arabinose (Fig 4c) 252

Intensification of rhTNFα production in fed-batch fermentations 253

The optimal conditions for the production of rhTNFα defined during shake-flask studies were 254

transferred to 5 L fed-batch fermentations using medium A a semi-defined culture medium 255

formulation obtained from Cobra Biologics supplemented with 2 casamino acids Production of 256

rhTNFα was induced with arabinose to a final concentration of 0005 at an OD600 of 05 after 3 257

hours of growth (Fig 5) The culture grew after inoculation reaching a final OD600 of 908 after 48 258

hours (Fig 5a) Although micro initially exceeded 04 during the initial batch phase of growth it 259

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

11

decreased below 04 after 6 hours The CFU dropped after 7 hours growth and plasmid retention 260

dropped below 90 after 11 hours (Fig 5b) After 48 hours only 2 of the bacteria had retained 261

the plasmid Although rhTNFα had accumulated to 22 of cellular protein after 26 h it did not 262

significantly increase after that point (Fig 5c) The increase in biomass between 26 h and 48 h did 263

not correlate with an increase in rhTNFα productivity as expected since the plasmid retention of the 264

culture was low and decreasing Production of the 31 kDa APH protein had also accumulated by 9 265

h post-induction Analysis of soluble and insoluble cellular fractions revealed that the majority of 266

rhTNFα was present in the soluble fraction (Fig 5d) 267

Taken together these data revealed that the fermentation conditions as defined in shake-flask 268

studies were a good starting point for fermentation development as rhTNFα was generated in the 269

soluble fraction in significant quantities and induction of rhTNFα production did not cause growth 270

arrest or immediate decreases in CFU or plasmid retention However harvesting the culture after 30 271

h when the percentage of cellular protein that was rhTNFα was the greatest (25 ) would not have 272

resulted in high overall rhTNFα yield due to the low biomass (OD600 lt 40) This indicates that 273

allocation of resources to growth and RPP was unbalanced In addition production of APH from 274

pLT72-TNFα could have increased the metabolic burden on the cells 275

Improvement of vector design for minimisation of APH synthesis 276

Although high yields of rhTNFα were obtained in bacteria transformed with plasmid pLT72-TNFα 277

plasmid deficient bacteria were detected 24 h post-induction under all growth conditions tested 278

Attempts were therefore made to decrease the stress on the host further by decreasing expression 279

of AHP which was suspected to be the abundant 31 kDa protein detected by SDS PAGE Three 280

modified plasmids were constructed (Fig 6a) and plasmid retention and rhTNFα accumulation 281

during growth in small scale batch cultures were compared with those of the original plasmid For 282

these experiments bacteria were grown at 30 degC in Terrific Broth and RP was induced with 002 283

arabinose These conditions were known from previous experiments to show high plasmid loss so 284

any improvement in plasmid retention would indicate reduced stress [6] 285

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

12

In contrast to the 20 plasmid retention by pLT72-TNFα transformant 40 of the bacteria 286

transformed with plasmid pLT72-T7t-TNFα in which the T7 terminator was cloned downstream of 287

the TNFα gene had retained the plasmid 24 h post-induction (Fig 6b) Similar results were obtained 288

with plasmid pLT72-T7tKanT2t-TNFα with both the T7 and the T2 terminators after the TNFα gene 289

and the kanamycin gene in reverse orientation However plasmid retention was further improved 290

(~80 ) in cultures carrying the vector with the T7 terminator sequence and the kanamycin 291

resistance gene in the opposite orientation to the rhTNFα gene (pLT72-T7tKan-TNFα) SDS-PAGE 292

analysis of whole cell proteins at different time points during growth revealed no significant 293

differences between the four vectors in terms of production of rhTNFα (Fig 6c) The abundance of 294

the 31 kDa protein was slightly lower with the vector containing the T7 terminator sequence (pLT72-295

T7t-TNFα) and decreased further for vectors with the sequence coding for kanamycin resistance 296

gene in reverse orientation (pLT72-T7tKan-TNFα and pLT72-T7tKanT2-TNFα) These data confirm 297

the identity of the 31 kDa protein as APH Plasmid retention data identified pLT72-T7tKan-TNFα as 298

the optimal construct due to its low rate of plasmid loss indicative of lowered stress 299

Improvement of fed-batch fermentation using the improved expression vector 300

The fed-batch fermentation was repeated with pLT72-T7tKan-TNFα containing the T7 terminator 301

sequence and the gene coding for the kanamycin resistance gene in reverse orientation 302

(Supplemental Fig S3) As before the culture grew continuously reaching a final OD600 of 68 after 303

48 h (Supplemental Fig S3a) although the OD600 only increased slightly after 33 h The specific 304

growth rate during the batch phase was comparable to the previous fermentation CFU values 305

decreased slightly less following induction (Supplemental Fig S3b) compared to the fermentation 306

using pLT72-TNFα (Fig 5a) whereas plasmid retention remained at close to 100 throughout 307

The quantity of rhTNFα was comparable to the fermentation using pLT72-TNFα with 25 of cellular 308

protein being rhTNFα after 27 h (Supplemental Fig S3c) After 48 h the majority (gt80 ) of rhTNFα 309

was in the soluble fraction The use of the improved vector design helped to minimise plasmid loss 310

during fermentation avoiding the overgrowth of plasmid-free cells and therefore non-productive 311

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

13

cells at the last stages of the fermentation process However this did not significantly increase the 312

quantity of rhTNFα generated 313

Removal of casamino acids from fed-batch growth medium 314

Good Manufacturing Practices (cGMP) for pharmaceutical products recommend that animal-derived 315

products should be excluded from bacterial culture medium to eliminate risks from zoonotic viruses 316

and transmissible spongiform encephalopathies To develop a GMP compliant fermentation process 317

casamino acids (a complex medium component derived from casein) were removed from medium A 318

and replaced with 14 gmiddotL-1 of ammonium sulphate and 03 gmiddotL-1 of calcium chloride Bacteria 319

transformed with plasmid pLT72-T7tKan-TNFα were induced at an OD600 of 05 with 0005 320

arabinose and sampled for up to 32 h post-induction The biomass concentration increased steadily 321

reaching a final OD600 of 72 (Supplemental Fig S4a) However no great increase of the cell biomass 322

was observed after 32 h and in contrast to previous fermentations the growth rate decreased 323

immediately following rhTNFα induction (Fig 5 amp Supplemental Fig S3) There was no decrease in 324

CFU following induction of RPP (Supplemental Fig S4b) but in contrast to 100 retention of the 325

plasmid in the presence of casamino acids plasmid-free bacteria started to overgrow the culture 326

after 9 h growth (Supplemental Fig S3) The concentration of rhTNFα increased after induction 327

reaching a maximum of 26 of total cell protein after 24 h growth (Supplemental Fig S4c) The 328

concentration of rhTNFα decreased after 24 h presumably due to outgrowth of plasmid-free non-329

productive cells The concentration of rhTNFα in the soluble fraction was lower than in previous 330

fermentations with only 50 of the rhTNFα in the soluble fraction after 48 h Thus removal of 331

casamino acids was detrimental to the fermentation process Ben-Bassat et al [15] and Hoffmann 332

et al [11] reported that casamino acids increased recombinant protein yields andor stability 333

presumably due to enhanced supply of amino acids allowing more rapid protein synthesis without 334

the need for generation of amino acids de novo Casamino acids contains more free amino acids 335

and more peptide fragments of smaller molecular mass (lt250 Da) than the yeast extract also present 336

in the medium [16] 337

Fermentation with an alternative medium composition 338

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

14

The semi-defined medium C is fully GMP compliant and has been used successfully both by 339

pharmaceutical companies and to produce model proteins in E coli fed-batch fermentations [7917] 340

This medium was therefore used in 5 L fed batch fermentations to generate rhTNFα from bacteria 341

transformed with plasmid pLT72-T7tKan-TNFα As before the addition of the feed was started 10 342

hours after inoculation and the specific growth rate (micro) during the fed batch phase was maintained 343

at 01 h-1 by the use of an exponential feeding profile The culture was induced at an OD600 of 05 by 344

the addition of arabinose to a final concentration of 0005 The pH was maintained at 70 by the 345

addition of 20 NH4OH or 5 M HCl 346

Exponential growth in medium C was faster than in media A or B peaking at micro = 0815 h-1 (Fig 7a) 347

Unlike previous fermentations the growth rate increased following induction of RPP then decreased 348

CFU measurements consistently increased throughout the fermentation (Fig 7b) and plasmid 349

retention remained at around 100 throughout The concentration of rhTNFα increased gradually 350

after induction reaching a maximum of 30 of the total cell protein between 24 and 30 h of growth 351

(Fig 7c) At the end of the fermentation rhTNFα constituted 25 of the total cell protein being 352

primarily accumulated in the soluble fraction with less than 30 of rhTNFα in the insoluble fraction 353

Overall this fermentation process successfully generated a greater yield of rhTNFα than previous 354

fermentations the majority being accumulated in a soluble form minimising plasmid loss and using 355

a culture medium compliant with cGMP guidelines 356

Bioassay of rhTNFα activity 357

The final goal of any fermentation process is the production of a soluble and active product and 358

protein solubility is often a good indication of activity However this statement is not always true and 359

recombinant proteins might accumulate in a soluble form but with an incorrect conformation which 360

limits or abolishes biological activity Therefore the activity of rhTNFα was measured using a 361

bioassay [18] rhTNFα was purified from the cell paste generated in fermentation 1 (Fig 5 purified 362

rhTNFα shown in Supplemental Fig S1) The C3H mouse fibrosarcoma cell line L929 which is 363

sensitive to hTNFα was used to evaluate the activity of the purified rhTNFα The basis of the 364

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

15

bioassay is that unlike dead cells killed by hTNFα live L929 cells are stained by the dye crystal violet 365

(CV) 366

The activity of the purified rhTNFα was calculated as the percentage of cytotoxicity by comparing 367

the amount of dye taken up by cells incubated with different quantities of rhTNFα rhTNFα samples 368

purified from fermentation 1 had a 50 cytotoxicity (LD50) value at a concentration of 0349 plusmn 0035 369

ng∙mL-1 This in within the concentration range of 005 to 20 ngmiddotmL-1 in which for most in vitro 370

applications hTNFα exerts its biological activity Comparison with LD50 values for rhTNFα standards 371

(Life Technologies) revealed that the specific activity of the purified rhTNFα was 28times106 Umiddotmg-1 372

Overall the result of the cytotoxicity assay showed that the optimization of the fermentation 373

conditions have led to the successful production of soluble and active product and the rhTNFα 374

produced during by fed-batch fermentation was active and stable 375

In summary stress minimisation has been demonstrated to be an effective tool for the optimisation 376

of the production of the human biopharmaceutical rhTNFα Data generated in shake-flask 377

experiments allowed design of intensified bioreactor cultures in which RPP and growth could be 378

balanced leading to high quantities of both rhTNFα and biomass Balanced growth allowing RP to 379

accumulate along with biomass and thus maintaining cell health and viability is important not only 380

for biomass and RP accumulation during the fermentation but also for harvest Unhealthy or 381

stressed bacteria are more difficult to harvest by centrifugation and subsequent downstream 382

processing steps [5]This strategy also enables more flexible scheduling of fermentations 383

384

AUTHOR STATEMENTS 385

Funding information This study was funded by Innovate UK the UK Biotechnology amp Biological 386

Sciences Research Council and the UK Engineering amp Physical Sciences Research Council under 387

the KTP scheme grant number KTP 9044 The funding bodies played no role in the design of the 388

study and collection analysis and interpretation of data or in writing the manuscript 389

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

16

Acknowledgements We thank Bruce Humphrey (Cobra Biologics) for plasmids and Nicola Barison 390

(Cobra Biologics) for assistance with protein purification 391

Conflicts of interest The authors declare that they have no competing financial interests 392

393

REFERENCES 394

1 Spadiut O Capone S Krainer F Glieder A Herwig C Microbials for the production of 395

monoclonal antibodies and antibody fragments Trends Biotechnol 2014 32 54-60 396

2 Overton TW Recombinant protein production in bacterial hosts Drug Discov Today 2014 19 397

590-601 398

3 Singh SM Panda AK Solubilization and refolding of bacterial inclusion body proteins J Biosci 399

Bioeng 2005 99 303ndash310 400

4 Soslashrensen HP Mortensen KK Soluble expression of recombinant proteins in the cytoplasm of 401

Escherichia coli Microb Cell Fact 200541 402

5 Hsu CC Thomas ORT Overton TW Periplasmic expression in and release of Fab fragments 403

from Escherichia coli using stress minimization J Chem Tech Biotechnol 2016 91815-822 404

6 Sevastsyanovich Y Alfasi S Overton T Hall R Jones J Hewitt C Cole J Exploitation of 405

GFP fusion proteins and stress avoidance as a generic strategy for the production of high‐quality 406

recombinant proteins FEMS Microbiol Lett 2009 299 86-94 407

7 Wyre C Overton TW Use of a stress-minimisation paradigm in high cell density fed-batch 408

Escherichia coli fermentations to optimise recombinant protein production J Ind Microbiol 409

Biotechnol 2014 41 1391-1404 410

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

17

8 Horiuchi T Mitoma H Harashima S Tsukamoto H Shimoda T Transmembrane TNF-α 411

structure function and interaction with anti-TNF agents Rheumatology (Oxford) 2010 491215-412

1228 413

9 Hodgson IJ Lennon CDJ Kara VL Expression system European patent EP2695943 (A1) 414

10 Oshima T Tanaka S Matsukura S Expression vector for human TNF US patent 415

US4871663 416

11 Hoffmann F van den Heuvel J Zidek N Rinas U Minimizing inclusion body formation during 417

recombinant protein production in Escherichia coli at bench and pilot plant scale Enz Microbial 418

Tech 2004 34235-241 419

12 Studier FW Use of bacteriophage T7 lysozyme to improve an inducible T7 expression system 420

J Mol Biol 1991 219 37-44 421

13 Sevastsyanovich YR Alfasi SN Cole JA Sense and nonsense from a systems biology 422

approach to microbial recombinant protein production Biotechnol Appl Biochem 2010 55 9-28 423

14 Alfasi SN Physiological aspects underpinning recombinant protein production in Escherichia 424

coli PhD Thesis University of Birmingham 2011 425

15 Ben-Bassat A Dorin G Bauer K Lin L Method of improving the yield of heterologous 426

protein produced by cultivating recombinant bacteria US patent US4656132 A 427

16 Beckton Dickinson and Company BD Bionutrients Technical Manual 4th Edition 428

httpwwwbdcomdstechnicalCentermisclcn01558-bionutrients-manualpdf Accessed on 7 June 429

2017 430

17 Want A Thomas OR Kara B Liddell J Hewitt CJ Studies related to antibody fragment 431

(Fab) production in Escherichia coli W3110 fed‐batch fermentation processes using multiparameter 432

flow cytometry Cytometry A 2009 75 148-154 433

18

18 McGee ZA Clemens CM Effect of bacterial products on tumor necrosis factor production 434

quantitation in biological fluids or tissues Methods Enzymol 1994 23623-31 435

19 Passarinha L Bonifacio M Queiroz J Application of a fed-batch bioprocess for the 436

heterologous production of hSCOMT in Escherichia coli J Microbiol Biotechnol 2009 19 972-981 437

20 Babaeipour V Shojaosadati S Robatjazi S Khalilzadeh R Maghsoudi N Over-production 438

of human interferon-γ by HCDC of recombinant Escherichia coli Process Biochem 2007 42112-439

117 440

FIGURE LEGENDS 441

Figure 1 Selection of culture medium for the optimisation of the production of rhTNFα E coli 442

BL21-T7 carrying the empty vector (pLT72) or the vector encoding rhTNFα (pLT72-TNFα) were 443

grown at 30 degC in LB (ace) or TB (bdf) half of cultures were induced with 02 arabinose at an 444

OD600 asymp 1 Samples were taken at regular intervals and (ab) OD600 (cd) CFU and (ef) plasmid 445

retention determined Data shown are single values for CFU and plasmid retention and mean values 446

from replica flasks for OD600 error bars are plusmn1 SD 447

Figure 2 Accumulation of rhTNFα in cultures grown in LB or TB culture media E coli BL21-448

T7 carrying the empty vector (pLT72) or the vector encoding rhTNFα (pLT72-TNFα) were grown in 449

(a) LB or (b) TB at 30 degC half of cultures were induced with 02 arabinose at an OD600 asymp 1 and 450

samples were taken at regular intervals Whole cell lysates were separated by SDS-PAGE and 451

protein stained with colloidal blue M marker BI before induction The quantity of rhTNFα is 452

expressed as a percentage of whole cell protein at the bottom of the gel Samples collected at 24 453

hours were also fractionated to obtain soluble (24S) and insoluble (24I) cell fractions facilitated by 454

the addition of lysozyme (shown on right) The ~31kDa protein presumed to be APH is shown on the 455

right 456

Figure 3 Effect of the inducer concentration on the production of rhTNFα (a) Growth of E coli 457

BL21-T7 carrying the empty vector (pLT72 non-induced) or pLT72-TNFα incubated at 30 degC and 458

induced with between 1 and 0002 arabinose at OD600 asymp 1 (b) Plasmid retention after 24 hours 459

19

of growth as in (a) (c) SDS-PAGE showing accumulation of rhTNFα from whole cell lysates obtained 460

24 hours post-inoculation Data shown are mean values from two replica flasks for OD600 and plasmid 461

retention error bars are plusmn1 standard deviation 462

Figure 4 Effect of the temperature and inducer concentration on the production of rhTNFα 463

(a) SDS-PAGE gel showing the accumulation of the rhTNFα in the soluble (Sol) and insoluble (Ins) 464

fractions 4 hours after induction Cultures were induced with 02 arabinose and grown at 37 degC 465

30 degC or 25 degC samples were fractionated and rhTNFα quantified by densitometry to give the 466

percentage of rhTNFα in the soluble and insoluble fractions (b) Plasmid retention of cultures carrying 467

either the empty vector (pLT72) or pLT72-TNFα were grown at 25 degC under non-induced conditions 468

or induced at OD600 asymp 05 with 02 to 0001 arabinose Data shown are mean values from two 469

replica flasks error bars are plusmn1 standard deviation (c) SDS-PAGE gel showing the accumulation of 470

rhTNFα from whole cell lysates after 24 hours of growth as in (b) 471

Figure 5 Production of rhTNFα by fed-batch fermentation E coli BL21-T7 pLT72-TNFα was 472

grown at 25 degC in medium A and induced with 0005 arabinose at OD600 asymp 05 (t = 3h solid arrow) 473

Feed was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific 474

growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 3h 475

sample is immediately before induction rhTNFα and APH are indicated along with densitometric 476

analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 477

pLT72 (empty vector) after 24 hours of growth (non-induced) (d) Samples were separated into 478

soluble and insoluble fractions rhTNFα and lysozyme are indicated 479

Figure 6 Optimisation of vector design to confirm the identity of the 31 kDa protein as APH 480

and to minimise its production (a) Schematic of the new vector designs Arrows indicate genes 481

stem-loops terminators (T7t and T2t) (b) Plasmid retention after 24 hours growth of cultures carrying 482

the empty vector (pLT72) or the 4 vectors as in (a) grown at 30 degC under non-inducer or induced 483

(002 arabinose at OD600 asymp 05) conditions Data shown are mean values from two replica flasks 484

error bars are plusmn1 standard deviation (c) SDS-PAGE gel showing the accumulation of rhTNFα and 485

APH in whole cell lysates after 4 h 8 h 10 h 12 h and 24 h growth as in (b) 486

20

Figure 7 Production of rhTNFα using an alternative fed-batch medium composition E coli 487

BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC in medium C and induced with 0005 arabinose 488

at an OD600 asymp 05 (t = 5h solid arrow) Feeding was started at t = 10h (dashed arrow) (a) Growth as 489

determined using OD600 and specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE 490

analysis of whole cell lysates The 5h sample is immediately before induction rhTNFα is indicated 491

along with densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo 492

sample is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 493

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to determine 494

the quantities of soluble and insoluble rhTNFα 495

496

21

FIGURE 1 497

498

499

500

501

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) LBpLT72 (induced) LBpLT72-TNFα (non-induced) LBpLT72-TNFα (induced) LB

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) TBpLT72 (induced) TBpLT72-TNFα (non-induced) TBpLT72-TNFα (induced) TB

a)

c)

e) f)

b)

d)

22

FIGURE 2 502

503

504

505

506

(kDa)

663 554

365 31

215

144

6

rhTNFα

M BI 35h 45h 55h 65h 24h 24S 24I 24h 24h 24h

pLT

72-T

NF

α

(Non

-indu

ced)

(kDa)

663 554

365 31

215

144

6

APH

Lysozyme

rhTNFα

a)

b)

cellular protein that is rhTNFα 5 10 13 15 20

cellular protein that is rhTNFα 5 11 12 15 20

APH

Lysozyme

Non

-indu

ced

Indu

ced

pLT72

pLT72-TNFα (Induced)

23

FIGURE 3 507

a) 508

509

510

511

512

513

514

515

516

517

b) 518

519

520

521

522

523

524

c) 525

526

527

528

529

530

531

532

533

534

0

20

40

60

80

100

pLT72 0 1 020 005 002 0002

P

lasm

id r

ete

nti

on

pLT72-TNFα

Arabinose concentration

(kDa)

28

17 14

M pLT72 0 1 02 005 002 0002

rhTNF

Arabinose concentration

pLT72-TNFα

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)

pLT72 (non-induced) pLT72-TNFα (non-induced)pLT72-TNFα (1) pLT72-TNFα (02)pLT72-TNFα (005) pLT72-TNFα (002)

pLT72-TNFα (0002)

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

19

of growth as in (a) (c) SDS-PAGE showing accumulation of rhTNFα from whole cell lysates obtained 460

24 hours post-inoculation Data shown are mean values from two replica flasks for OD600 and plasmid 461

retention error bars are plusmn1 standard deviation 462

Figure 4 Effect of the temperature and inducer concentration on the production of rhTNFα 463

(a) SDS-PAGE gel showing the accumulation of the rhTNFα in the soluble (Sol) and insoluble (Ins) 464

fractions 4 hours after induction Cultures were induced with 02 arabinose and grown at 37 degC 465

30 degC or 25 degC samples were fractionated and rhTNFα quantified by densitometry to give the 466

percentage of rhTNFα in the soluble and insoluble fractions (b) Plasmid retention of cultures carrying 467

either the empty vector (pLT72) or pLT72-TNFα were grown at 25 degC under non-induced conditions 468

or induced at OD600 asymp 05 with 02 to 0001 arabinose Data shown are mean values from two 469

replica flasks error bars are plusmn1 standard deviation (c) SDS-PAGE gel showing the accumulation of 470

rhTNFα from whole cell lysates after 24 hours of growth as in (b) 471

Figure 5 Production of rhTNFα by fed-batch fermentation E coli BL21-T7 pLT72-TNFα was 472

grown at 25 degC in medium A and induced with 0005 arabinose at OD600 asymp 05 (t = 3h solid arrow) 473

Feed was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific 474

growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 3h 475

sample is immediately before induction rhTNFα and APH are indicated along with densitometric 476

analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 477

pLT72 (empty vector) after 24 hours of growth (non-induced) (d) Samples were separated into 478

soluble and insoluble fractions rhTNFα and lysozyme are indicated 479

Figure 6 Optimisation of vector design to confirm the identity of the 31 kDa protein as APH 480

and to minimise its production (a) Schematic of the new vector designs Arrows indicate genes 481

stem-loops terminators (T7t and T2t) (b) Plasmid retention after 24 hours growth of cultures carrying 482

the empty vector (pLT72) or the 4 vectors as in (a) grown at 30 degC under non-inducer or induced 483

(002 arabinose at OD600 asymp 05) conditions Data shown are mean values from two replica flasks 484

error bars are plusmn1 standard deviation (c) SDS-PAGE gel showing the accumulation of rhTNFα and 485

APH in whole cell lysates after 4 h 8 h 10 h 12 h and 24 h growth as in (b) 486

20

Figure 7 Production of rhTNFα using an alternative fed-batch medium composition E coli 487

BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC in medium C and induced with 0005 arabinose 488

at an OD600 asymp 05 (t = 5h solid arrow) Feeding was started at t = 10h (dashed arrow) (a) Growth as 489

determined using OD600 and specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE 490

analysis of whole cell lysates The 5h sample is immediately before induction rhTNFα is indicated 491

along with densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo 492

sample is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 493

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to determine 494

the quantities of soluble and insoluble rhTNFα 495

496

21

FIGURE 1 497

498

499

500

501

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) LBpLT72 (induced) LBpLT72-TNFα (non-induced) LBpLT72-TNFα (induced) LB

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) TBpLT72 (induced) TBpLT72-TNFα (non-induced) TBpLT72-TNFα (induced) TB

a)

c)

e) f)

b)

d)

22

FIGURE 2 502

503

504

505

506

(kDa)

663 554

365 31

215

144

6

rhTNFα

M BI 35h 45h 55h 65h 24h 24S 24I 24h 24h 24h

pLT

72-T

NF

α

(Non

-indu

ced)

(kDa)

663 554

365 31

215

144

6

APH

Lysozyme

rhTNFα

a)

b)

cellular protein that is rhTNFα 5 10 13 15 20

cellular protein that is rhTNFα 5 11 12 15 20

APH

Lysozyme

Non

-indu

ced

Indu

ced

pLT72

pLT72-TNFα (Induced)

23

FIGURE 3 507

a) 508

509

510

511

512

513

514

515

516

517

b) 518

519

520

521

522

523

524

c) 525

526

527

528

529

530

531

532

533

534

0

20

40

60

80

100

pLT72 0 1 020 005 002 0002

P

lasm

id r

ete

nti

on

pLT72-TNFα

Arabinose concentration

(kDa)

28

17 14

M pLT72 0 1 02 005 002 0002

rhTNF

Arabinose concentration

pLT72-TNFα

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)

pLT72 (non-induced) pLT72-TNFα (non-induced)pLT72-TNFα (1) pLT72-TNFα (02)pLT72-TNFα (005) pLT72-TNFα (002)

pLT72-TNFα (0002)

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

20

Figure 7 Production of rhTNFα using an alternative fed-batch medium composition E coli 487

BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC in medium C and induced with 0005 arabinose 488

at an OD600 asymp 05 (t = 5h solid arrow) Feeding was started at t = 10h (dashed arrow) (a) Growth as 489

determined using OD600 and specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE 490

analysis of whole cell lysates The 5h sample is immediately before induction rhTNFα is indicated 491

along with densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo 492

sample is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 493

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to determine 494

the quantities of soluble and insoluble rhTNFα 495

496

21

FIGURE 1 497

498

499

500

501

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) LBpLT72 (induced) LBpLT72-TNFα (non-induced) LBpLT72-TNFα (induced) LB

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) TBpLT72 (induced) TBpLT72-TNFα (non-induced) TBpLT72-TNFα (induced) TB

a)

c)

e) f)

b)

d)

22

FIGURE 2 502

503

504

505

506

(kDa)

663 554

365 31

215

144

6

rhTNFα

M BI 35h 45h 55h 65h 24h 24S 24I 24h 24h 24h

pLT

72-T

NF

α

(Non

-indu

ced)

(kDa)

663 554

365 31

215

144

6

APH

Lysozyme

rhTNFα

a)

b)

cellular protein that is rhTNFα 5 10 13 15 20

cellular protein that is rhTNFα 5 11 12 15 20

APH

Lysozyme

Non

-indu

ced

Indu

ced

pLT72

pLT72-TNFα (Induced)

23

FIGURE 3 507

a) 508

509

510

511

512

513

514

515

516

517

b) 518

519

520

521

522

523

524

c) 525

526

527

528

529

530

531

532

533

534

0

20

40

60

80

100

pLT72 0 1 020 005 002 0002

P

lasm

id r

ete

nti

on

pLT72-TNFα

Arabinose concentration

(kDa)

28

17 14

M pLT72 0 1 02 005 002 0002

rhTNF

Arabinose concentration

pLT72-TNFα

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)

pLT72 (non-induced) pLT72-TNFα (non-induced)pLT72-TNFα (1) pLT72-TNFα (02)pLT72-TNFα (005) pLT72-TNFα (002)

pLT72-TNFα (0002)

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

21

FIGURE 1 497

498

499

500

501

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) LBpLT72 (induced) LBpLT72-TNFα (non-induced) LBpLT72-TNFα (induced) LB

0

2

4

6

8

10

12

14

16

18

0 5 10 15 20 25

OD

600

Time (h)

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 5 10 15 20 25

CF

Um

L-1

Time (h)

0

20

40

60

80

100

0 5 10 15 20 25

Pla

sm

id r

ete

nti

on

(

)

Time (h)

pLT72 (non-induced) TBpLT72 (induced) TBpLT72-TNFα (non-induced) TBpLT72-TNFα (induced) TB

a)

c)

e) f)

b)

d)

22

FIGURE 2 502

503

504

505

506

(kDa)

663 554

365 31

215

144

6

rhTNFα

M BI 35h 45h 55h 65h 24h 24S 24I 24h 24h 24h

pLT

72-T

NF

α

(Non

-indu

ced)

(kDa)

663 554

365 31

215

144

6

APH

Lysozyme

rhTNFα

a)

b)

cellular protein that is rhTNFα 5 10 13 15 20

cellular protein that is rhTNFα 5 11 12 15 20

APH

Lysozyme

Non

-indu

ced

Indu

ced

pLT72

pLT72-TNFα (Induced)

23

FIGURE 3 507

a) 508

509

510

511

512

513

514

515

516

517

b) 518

519

520

521

522

523

524

c) 525

526

527

528

529

530

531

532

533

534

0

20

40

60

80

100

pLT72 0 1 020 005 002 0002

P

lasm

id r

ete

nti

on

pLT72-TNFα

Arabinose concentration

(kDa)

28

17 14

M pLT72 0 1 02 005 002 0002

rhTNF

Arabinose concentration

pLT72-TNFα

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)

pLT72 (non-induced) pLT72-TNFα (non-induced)pLT72-TNFα (1) pLT72-TNFα (02)pLT72-TNFα (005) pLT72-TNFα (002)

pLT72-TNFα (0002)

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

22

FIGURE 2 502

503

504

505

506

(kDa)

663 554

365 31

215

144

6

rhTNFα

M BI 35h 45h 55h 65h 24h 24S 24I 24h 24h 24h

pLT

72-T

NF

α

(Non

-indu

ced)

(kDa)

663 554

365 31

215

144

6

APH

Lysozyme

rhTNFα

a)

b)

cellular protein that is rhTNFα 5 10 13 15 20

cellular protein that is rhTNFα 5 11 12 15 20

APH

Lysozyme

Non

-indu

ced

Indu

ced

pLT72

pLT72-TNFα (Induced)

23

FIGURE 3 507

a) 508

509

510

511

512

513

514

515

516

517

b) 518

519

520

521

522

523

524

c) 525

526

527

528

529

530

531

532

533

534

0

20

40

60

80

100

pLT72 0 1 020 005 002 0002

P

lasm

id r

ete

nti

on

pLT72-TNFα

Arabinose concentration

(kDa)

28

17 14

M pLT72 0 1 02 005 002 0002

rhTNF

Arabinose concentration

pLT72-TNFα

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)

pLT72 (non-induced) pLT72-TNFα (non-induced)pLT72-TNFα (1) pLT72-TNFα (02)pLT72-TNFα (005) pLT72-TNFα (002)

pLT72-TNFα (0002)

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

23

FIGURE 3 507

a) 508

509

510

511

512

513

514

515

516

517

b) 518

519

520

521

522

523

524

c) 525

526

527

528

529

530

531

532

533

534

0

20

40

60

80

100

pLT72 0 1 020 005 002 0002

P

lasm

id r

ete

nti

on

pLT72-TNFα

Arabinose concentration

(kDa)

28

17 14

M pLT72 0 1 02 005 002 0002

rhTNF

Arabinose concentration

pLT72-TNFα

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)

pLT72 (non-induced) pLT72-TNFα (non-induced)pLT72-TNFα (1) pLT72-TNFα (02)pLT72-TNFα (005) pLT72-TNFα (002)

pLT72-TNFα (0002)

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

24

FIGURE 4 535

536

537

538

0

20

40

60

80

100

pLT72 0 020 002 0005 0002 0001

P

lasm

id r

ete

nti

on

M Sol Ins Sol Ins Sol Ins

37 degC 30 degC 25 degC

(kDa)

28

17

14

rhTNF

α

a)

pLT72-TNFα

arabinose concentration

M pLT72 0 0005 0002 0001 (kDa)

28

17 14

rhTNF

α

c)

54

soluble

73 soluble 90 soluble

pLT72-TNFα - arabinose concentration

b)

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

25

FIGURE 5 539

540

541

Time post-inoculation

3h 11h 24h 32h 48h

M Sol Ins Sol Ins Sol Ins Sol Ins Sol Ins -ve (kDa)

28

17 14

M 3h 5h 9h 11h 24h 26h 28h 30h 32h 48h -ve

(kDa)

188

98

62

49

38

28

17 14

6

APH

rhTNFα

Lysozyme

c)

d)

4 10 15 20 22 22 25 22 20

cellular protein that is rhTNFα

000

010

020

030

040

050

0

20

40

60

80

100

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

micro

(h-1

)

OD

600

OD600

Specific growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

rhTNFα

a)

b)

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

26

0

20

40

60

80

100

120

P

lasm

id r

ete

nti

on

FIGURE 6 542

543

544

545

546

547

548

hTNFα Kanamycin pLT72-TNFα

pLT72-T7t-TNFα

pLT72-T7tKan-TNFα

pLT72-T7tKanT2t-TNFα

pT7

pT7

pT7

pT7

hTNFα

hTNFα

hTNFα

Kanamycin

Kanamycin

Kanamycin

pLT

72

(Non

-

indu

ced)

pL

T72

(Ind

uced

)

pLT

72

(Non

-ind

) pL

T72

(Ind

uced

)

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-TNFα pLT72-T7t-TNFα

M 4h 8h 10h 12h M 4h 8h 10h 12h 24h 24h

pLT72-T7tKan-TNFα pLT72-T7tT7tKanT2t-TNFα

(kDa)

38

28

17 14

(kDa)

38

28

17 14

rhTNFα

rhTNFα

APH

APH

c)

Non-induced Induced (002)

T7t

T7t

T7t

T2t

a)

b)

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

27

FIGURE 7 549

550

551

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17 14

Time post-inoculation Sol Ins

16 26 30 30 27 25

cellular protein that is rhTNFα

0

02

04

06

08

1

0

20

40

60

80

100

120

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

μ(h

-1)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

a)

b)

c)

rhTNFα

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

28

Supplemental information for 552

Optimising host cell physiology and stress avoidance for the production of 553

recombinant human tumour necrosis factor α in Escherichia coli 554

T Selas Castintildeeiras123 SG Williams1 A Hitchcock1 JA Cole34 DC Smith1 TW 555

Overton23 556

1Cobra Biologics Stephenson Building The Science Park Keele ST5 5SP UK 557

2School of Chemical Engineering 3Institute of Microbiology amp Infection and 4School of 558

Biosciences The University of Birmingham Edgbaston Birmingham B15 2TT UK 559

To whom correspondence should be sent twovertonbhamacuk 560

561

562

563

29

564

565

566

567

568

569

570

571

572

573

574

Supplemental Figure S1 Final material obtained after the purification process of rhTNFα The 575

rhTNFα accumulated mainly in the monomeric form The presence of the rhTNFα dimer and trimer 576

could also be detected by SDS-PAGE (left) and western blot (right) Bands corresponding to other 577

contaminant proteins could also be observed (orange circles) The final product had a purity greater 578

than 95 as determined by densitometry The western blot was developed using an anti-TNFα 579

antibody which confirmed the identity of the monomer dimer and trimer forms of the rhTNFα 580

produced by fed-batch fermentation 581

582

(kDa)

188

98

62

49

38

28

17 14

6

Trimer

Dimer

Monomer

(kDa) 188 98 62 49 38 28 17 14 6

30

SUPPLEMENTAL FIGURE S2 583

a) 584

585

586

587

588

589

590

591

592

b) 593

594

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)pLT72-TNFα pLT72 (non-induced) pLT72-TNFα OD600=05

pLT72-TNFα OD600=2 pLT72-TNFα OD600=3

M BI 2h 4h BI 2h 4h BI 2h 4h 24h 24h (kDa)

663

554

365

31

215

144

6

25

rhTNFα

pLT

72 (

Non

-induce

d)

pLT

72-T

NF

α

(Non-in

duced)

OD600=2 OD600=0

5

OD600=3

(non-induced)

31

Supplemental Figure S2 The effect of the induction point on the production of rhTNFα (a) 595

Growth of E coli BL21-T7 carrying the empty vector (pLT72) or the vector coding for rhTNFα (pLT72-596

TNFα) incubated at 30 degC cultures were induced with 02 arabinose at OD600 asymp 05 2 or 3 (b) 597

SDS-PAGE showing accumulation of rhTNFα from whole cell lysates before induction (BI) 2 hours 598

and 4 hours after induction Data shown are mean values from two replica flasks for OD600 error 599

bars are plusmn1 standard deviation 600

601

32

SUPPLEMENTAL FIGURE S3 602

603

604

605

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

a)

b)

Time post-inoculation)

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

6

rhTNFα

15 21 23 25 25 25 20

cellular protein that is rhTNFα

Sol Ins c)

33

Supplemental Figure S3 Production of rhTNFα by fed-batch fermentation using optimised 606

expression vector pLT72-T7tKan-TNFα E coli BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC 607

in medium A and induced with 0005 arabinose at an OD600 asymp 05 (t = 5h solid arrow) Feeding 608

was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific growth 609

rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 5h sample is 610

immediately before induction rhTNFα is indicated along with densitometric analysis of the 611

percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 pLT72 (empty 612

vector non-induced) after 24 hours of growth Samples obtained at t = 48h were also fractionated 613

into soluble (Sol) and insoluble (Ins) fractions to determine the quantities of soluble and insoluble 614

rhTNFα 615

616

34

SUPPLEMENTAL FIGURE S4 617

618

619

620

621

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFUmL

Plasmid retention

M 3h 5h 9h 24h 28h 30h 32h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

Time post-inoculation) Sol Ins

25 26 20 20 20 18

cellular protein that is rhTNFα

rhTNFα

a)

c)

b)

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

30

SUPPLEMENTAL FIGURE S2 583

a) 584

585

586

587

588

589

590

591

592

b) 593

594

0

2

4

6

8

10

12

14

16

0 5 10 15 20 25

OD

600

Time (h)pLT72-TNFα pLT72 (non-induced) pLT72-TNFα OD600=05

pLT72-TNFα OD600=2 pLT72-TNFα OD600=3

M BI 2h 4h BI 2h 4h BI 2h 4h 24h 24h (kDa)

663

554

365

31

215

144

6

25

rhTNFα

pLT

72 (

Non

-induce

d)

pLT

72-T

NF

α

(Non-in

duced)

OD600=2 OD600=0

5

OD600=3

(non-induced)

31

Supplemental Figure S2 The effect of the induction point on the production of rhTNFα (a) 595

Growth of E coli BL21-T7 carrying the empty vector (pLT72) or the vector coding for rhTNFα (pLT72-596

TNFα) incubated at 30 degC cultures were induced with 02 arabinose at OD600 asymp 05 2 or 3 (b) 597

SDS-PAGE showing accumulation of rhTNFα from whole cell lysates before induction (BI) 2 hours 598

and 4 hours after induction Data shown are mean values from two replica flasks for OD600 error 599

bars are plusmn1 standard deviation 600

601

32

SUPPLEMENTAL FIGURE S3 602

603

604

605

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

a)

b)

Time post-inoculation)

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

6

rhTNFα

15 21 23 25 25 25 20

cellular protein that is rhTNFα

Sol Ins c)

33

Supplemental Figure S3 Production of rhTNFα by fed-batch fermentation using optimised 606

expression vector pLT72-T7tKan-TNFα E coli BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC 607

in medium A and induced with 0005 arabinose at an OD600 asymp 05 (t = 5h solid arrow) Feeding 608

was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific growth 609

rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 5h sample is 610

immediately before induction rhTNFα is indicated along with densitometric analysis of the 611

percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 pLT72 (empty 612

vector non-induced) after 24 hours of growth Samples obtained at t = 48h were also fractionated 613

into soluble (Sol) and insoluble (Ins) fractions to determine the quantities of soluble and insoluble 614

rhTNFα 615

616

34

SUPPLEMENTAL FIGURE S4 617

618

619

620

621

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFUmL

Plasmid retention

M 3h 5h 9h 24h 28h 30h 32h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

Time post-inoculation) Sol Ins

25 26 20 20 20 18

cellular protein that is rhTNFα

rhTNFα

a)

c)

b)

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

31

Supplemental Figure S2 The effect of the induction point on the production of rhTNFα (a) 595

Growth of E coli BL21-T7 carrying the empty vector (pLT72) or the vector coding for rhTNFα (pLT72-596

TNFα) incubated at 30 degC cultures were induced with 02 arabinose at OD600 asymp 05 2 or 3 (b) 597

SDS-PAGE showing accumulation of rhTNFα from whole cell lysates before induction (BI) 2 hours 598

and 4 hours after induction Data shown are mean values from two replica flasks for OD600 error 599

bars are plusmn1 standard deviation 600

601

32

SUPPLEMENTAL FIGURE S3 602

603

604

605

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

a)

b)

Time post-inoculation)

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

6

rhTNFα

15 21 23 25 25 25 20

cellular protein that is rhTNFα

Sol Ins c)

33

Supplemental Figure S3 Production of rhTNFα by fed-batch fermentation using optimised 606

expression vector pLT72-T7tKan-TNFα E coli BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC 607

in medium A and induced with 0005 arabinose at an OD600 asymp 05 (t = 5h solid arrow) Feeding 608

was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific growth 609

rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 5h sample is 610

immediately before induction rhTNFα is indicated along with densitometric analysis of the 611

percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 pLT72 (empty 612

vector non-induced) after 24 hours of growth Samples obtained at t = 48h were also fractionated 613

into soluble (Sol) and insoluble (Ins) fractions to determine the quantities of soluble and insoluble 614

rhTNFα 615

616

34

SUPPLEMENTAL FIGURE S4 617

618

619

620

621

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFUmL

Plasmid retention

M 3h 5h 9h 24h 28h 30h 32h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

Time post-inoculation) Sol Ins

25 26 20 20 20 18

cellular protein that is rhTNFα

rhTNFα

a)

c)

b)

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

32

SUPPLEMENTAL FIGURE S3 602

603

604

605

0

20

40

60

80

100

1E+06

1E+07

1E+08

1E+09

1E+10

1E+11

1E+12

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFU mL-1

Plasmid retention

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50

Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

a)

b)

Time post-inoculation)

M 5h 9h 11h 24h 27h 30h 33h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

6

rhTNFα

15 21 23 25 25 25 20

cellular protein that is rhTNFα

Sol Ins c)

33

Supplemental Figure S3 Production of rhTNFα by fed-batch fermentation using optimised 606

expression vector pLT72-T7tKan-TNFα E coli BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC 607

in medium A and induced with 0005 arabinose at an OD600 asymp 05 (t = 5h solid arrow) Feeding 608

was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific growth 609

rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 5h sample is 610

immediately before induction rhTNFα is indicated along with densitometric analysis of the 611

percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 pLT72 (empty 612

vector non-induced) after 24 hours of growth Samples obtained at t = 48h were also fractionated 613

into soluble (Sol) and insoluble (Ins) fractions to determine the quantities of soluble and insoluble 614

rhTNFα 615

616

34

SUPPLEMENTAL FIGURE S4 617

618

619

620

621

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFUmL

Plasmid retention

M 3h 5h 9h 24h 28h 30h 32h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

Time post-inoculation) Sol Ins

25 26 20 20 20 18

cellular protein that is rhTNFα

rhTNFα

a)

c)

b)

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

33

Supplemental Figure S3 Production of rhTNFα by fed-batch fermentation using optimised 606

expression vector pLT72-T7tKan-TNFα E coli BL21-T7 pLT72-T7tKan-TNFα was grown at 25 degC 607

in medium A and induced with 0005 arabinose at an OD600 asymp 05 (t = 5h solid arrow) Feeding 608

was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and specific growth 609

rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell lysates The 5h sample is 610

immediately before induction rhTNFα is indicated along with densitometric analysis of the 611

percentage of cellular protein that is rhTNFα The ldquondashverdquo sample is E coli BL21-T7 pLT72 (empty 612

vector non-induced) after 24 hours of growth Samples obtained at t = 48h were also fractionated 613

into soluble (Sol) and insoluble (Ins) fractions to determine the quantities of soluble and insoluble 614

rhTNFα 615

616

34

SUPPLEMENTAL FIGURE S4 617

618

619

620

621

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFUmL

Plasmid retention

M 3h 5h 9h 24h 28h 30h 32h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

Time post-inoculation) Sol Ins

25 26 20 20 20 18

cellular protein that is rhTNFα

rhTNFα

a)

c)

b)

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

34

SUPPLEMENTAL FIGURE S4 617

618

619

620

621

0

01

02

03

04

05

06

0

10

20

30

40

50

60

70

80

0 10 20 30 40 50 Sp

ecif

ic g

row

th r

ate

(μ

)

OD

600

OD600

Growth rate

0

20

40

60

80

100

1E+07

1E+08

1E+09

1E+10

1E+11

0 10 20 30 40 50

P

lasm

id r

ete

nti

on

CF

U m

L-1

Time (h)

CFUmL

Plasmid retention

M 3h 5h 9h 24h 28h 30h 32h 48h 48h 48h -ve (kDa)

188

98

62

49

38

28

17

14

Time post-inoculation) Sol Ins

25 26 20 20 20 18

cellular protein that is rhTNFα

rhTNFα

a)

c)

b)

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

35

Supplemental Figure S4 Production of rhTNFα by fed-batch fermentation without the 622

addition of casamino acids E coli BL21-T7 pLT72-T7tKan-TNFα was grown in medium 623

B at 25 degC and induced with 0005 arabinose at an OD600 asymp 05 (t = 3h solid arrow) 624

Feeding was started at t = 10h (dashed arrow) (a) Growth as determined using OD600 and 625

specific growth rate (b) CFU and plasmid retention (c) SDS-PAGE analysis of whole cell 626

lysates The 3h sample is immediately before induction rhTNFα is indicated along with 627

densitometric analysis of the percentage of cellular protein that is rhTNFα The ldquondashverdquo sample 628

is E coli BL21-T7 pLT72 (empty vector non-induced) after 24 hours of growth Samples 629

obtained at t = 48h were also fractionated into soluble (Sol) and insoluble (Ins) fractions to 630

determine the quantities of soluble and insoluble rhTNFα 631

632

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

36

Supplemental methods 633

Fed-batch fermentation 634

Starter cultures were grown in 10 mL of LB broth with 50 μgmiddotmL-1 kanamycin at 25 degC and 635

200 rpm until OD600 = 2 Starter cultures were used to inoculate a 1 L baffled shake-flask 636

containing 200 mL of LB broth with 50 mgmiddotL-1 kanamycin and were grown at 25 degC and 200 637

rpm to an OD600 between 4 and 6 638

A 7 L total volume (5 L working volume) bench-top fermenter (Applikon ADI 1010 Bio 639

controller) equipped with 3 Rushton impellers and 4 baffles was used for fermentation 640

experiments The aeration rate was constant at 1 volume air per volume medium per minute 641

(vvm) and the dissolved oxygen tension (DOT) was maintained above 20 being controlled 642

by the stirrer speed (200 - 1250 rpm) All fermentations began with an initial volume of 3 L 643

batch salts sterilised in the vessel by autoclaving for 20 minutes at 121 degC Once cooled 644

post-autoclave additions and trace element solutions were added Two litres of feed solution 645

was prepared and sterilised by filtration (022 microm filter) For medium A (Cobra biologics) the 646

batch salts contained 133 gmiddotL-1 K2HPO4 4 gmiddotL-1 (NH4)2SO4 17 gmiddotL-1 citric acid 10 gmiddotL-1 647

BactoTM yeast extract and 016 mLmiddotL-1 PPG 2000 The post-autoclave additions were 1 648

mLmiddotL-1 trace elements solution A (comprising 5 gmiddotL-1 citric acid 2 gmiddotL-1 CoCl2middot6H2O 12 gmiddotL-649

1 CuCl2middot2H2O 25 gmiddotL-1 H3BO3 2 gmiddotL-1 Na2MoO4middot2H2O 12 gmiddotL-1 MnCl2middot4H2O) 10 mLmiddotL-1 650

trace elements solution B (comprising 6 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 EDTAmiddot2H2O and 08 651

gmiddotL-1 ZnCl2) 10 gmiddotL-1 glycerol 12 gmiddotL-1 MgSO4middot7H2O 2 (wv) casamino acids and 1 mLmiddotL-652

1 50 mg middot mL-1 kanamycin stock The feed solution contained 600 gmiddotL-1 glycerol 5 gmiddotL-1 653

MgSO4middot7H2O 50 gmiddotL-1 yeast extract 10 gmiddotL-1 KH2PO4 21 gmiddotL-1 K2HPO4 2 (wv) casamino 654

acids 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-1 20 arabinose stock For 655

medium B casamino acids were omitted and replaced with 14 gmiddotL-1 (NH4)2SO4 and 03 gmiddotL-656

1 of CaCl2middot2H2O 657

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

37

For medium C [17] the batch salts contained 14 gmiddotL-1 (NH4)2SO4 35 gmiddotL-1 glycerol 20 gmiddotL-1 658

BactoTM yeast extract 2 gmiddotL-1 KH2PO4 165 gmiddotL-1 K2HPO4 75 gmiddotL-1 citric acid 15 mL middot L-1 659

concentrated H3PO4 and 066 mLmiddotL-1 PPG 2000 The post-autoclave additions were 34 660

mLmiddotL-1 trace elements solution (comprising 336 gmiddotL-1 FeSO4middot7H2O 084 gmiddotL-1 ZnSO4middot7H2O 661

015 gmiddotL-1 MnSO4middotH2O 025 gmiddotL-1 Na2MoO4middot2H2O 012 gmiddotL-1 CuSO4middot5H2O 036 gmiddotL-1 662

H3BO3 and 48 mLmiddotL-1 concentrated H3PO4) 10 mLmiddotL-1 1 M MgSO4middot7H2O 2 mLmiddotL-1 1 M 663

CaCl2middot2H2O and 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock The feed contained 714 gmiddotL-1 664

glycerol 30 mLmiddotL-1 1 M MgSO4middot7H2O 1 mLmiddotL-1 50 mgmiddotmL-1 kanamycin stock and 05 mLmiddotL-665

1 20 arabinose stock 666

The pH was maintained at 68 by the addition of 5 M NaOH and 5 M HCl for fermentations 667

using media A and B and at 70 by the addition of 5 M HCl or 20 NH4OH for medium C 668

Polypropylene glycol (PPG) antifoam was added when required Fed-batch fermentations 669

were monitored using BioXpertreg software (Applikon) The inoculum was added to an initial 670

OD600 of 01 The fermentation was carried out at a temperature of 25 degC and the culture 671

was induced with 0005 arabinose at an OD600 of 05 The feed solution was started 10 h 672

after inoculation at an exponential feed rate to achieve a specific growth rate of 01 h-1 673

calculated using equation 1 674

119865 = (1

119878) times (

120583

119884119883119878+ 119898) times 1198830 times 119890120583119905 (1) 675

F is the feed rate in Lmiddoth-1 S is the substrate concentration in the feed in gmiddotL-1 μ is the required 676

specific growth rate in h-1 YXS is the yield coefficient in g biomass per g carbon source m is 677

the maintenance coefficient X0 is the biomass in g and t is time Values for YXS and m were 678

obtained from the literature 022 [19] and 0025 [20] respectively 679

680

681

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

38

SDS-PAGE 682

4-12 Bis-Tris NuPAGE SDS-PAGE gels (Life Technologies) were generally used to 683

evaluate the production of recombinant proteins Seven microliters of protein sample were 684

mixed with 2 μL of 4x NuPAGE LDS sample buffer (Life Technologies) and 1 μL of 10x 685

NuPAGE sample reducing agent (Life Technologies) and heated for 10 min at 70 degC 1x 686

electrophoresis running buffer was prepared by diluting 20x NuPAGE MES SDS running 687

buffer (Life Technologies) in deionised water For reducing protein electrophoresis 05 mL 688

of NuPAGE antioxidant (Life Technologies) was added to 200 mL of running buffer and used 689

to fill the inner chamber of the electrophoresis tank SDS-PAGE gels were run for at least 690

45 minutes at 200 V according to the manufacturersrsquo protocol Molecular size markers were 691

used Mark12trade Unstained Protein Standard (Life Technologies) or SeeBluereg Plus2 Pre-692

Stained Protein Standard (Life Technologies) SDS-PAGE gels were stained using Colloidal 693

Blue Staining (Life Technologies) SDS-PAGE gels were submerged in the fixing solution 694

(40 (vv) methanol 10 (vv) glacial acetic acid) for 10 minutes the staining solution A 695

(20 (vv) methanol and 20 (vv) staining solution A) for 10 minutes then staining solution 696

B was added to a final concentration of 5 (vv) SDS-PAGE gels were stained for a 697

minimum of 3 hours and de-stained with deionised water for at least 12 hours 698

Western blotting 699

SDS-PAGE gels were run as above and transferred to a 02 μm nitrocellulose membrane 700

(Life Technologies) using the Xcell II blot module at 30 V for 1 h (Life Technologies) Transfer 701

buffer was prepared by the addition of 20x NuPAGE transfer buffer (Life Technologies) 10 702

of methanol (vv) and 1 mLmiddotL-1 NuPAGE antioxidant Membranes were blocked in 5 (wv) 703

skimmed milk powder (Sigma-Aldrich) in PBS for at least 1 hour For the detection of 704

rhTNFα the blot was incubated with an anti-TNFα antibody (ab9635 Abcam) using 12500 705

dilution in 5 (wv) skimmed milk (Sigma-Aldrich) in PBS for 1 h washed with 005 Tween-706

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727

39

20 in PBS and successively incubated with an anti-rabbit IgG antibody conjugated with 707

horseradish peroxidase (HRP Sigma-Aldrich) using 13000 dilution for an hour Western 708

blots were developed using 33prime55prime-tetramethylbenzidine (TMB) substrate for HRP (Sigma-709

Aldrich) 710

Quantification of rhTNFα from SDS-PAGE 711

AlphaEasereg software (Alpha Innotech) was used to calculate the quantity of rhTNFα as a 712

percentage of total cell protein (TCP) Gels were photographed using an AlphaImager 713

(Alpha Innotech) and images subjected to background subtraction using the default settings 714

for peak-to-peak background subtraction The percentage of soluble and insoluble 715

recombinant protein was calculated by the software package All samples were normalised 716

by OD600 before loading on the SDS-PAGE gel so each lane contained equivalent biomass 717

rhTNFα reference material obtained from Life Technologies was used to quantify the 718

concentration of rhTNFα obtained at the end of each fermentation by densitometry from 719

samples A standard curve was generated by loading different concentrations of rhTNFα 720

reference material on an SDS-PAGE gel The concentration of rhTNFα from fermentation 721

samples was quantified using a standard curve with the AlphaEasereg software The rhTNFα 722

yields were calculated to obtain the final yield based on the final OD600 of the culture 723

724

725

726

727